ZHIVEXzhivex_ai
Python 0.25.0 · published wheel reference and guides.
PythonMixed
Public import paths and stability come from the installed artifact.
Use the root for Stable APIs and focused namespaces for extensions. Legacy root extension imports remain classified below.
A2AAgentCard
Beta · protocol
from zhivex_ai import A2AAgentCardA2AAgentCard(name: 'str', description: 'str', url: 'str', version: 'str', skills: 'list[A2AAgentSkill]', protocol_version: 'str' = '1.0', preferred_transport: 'str' = 'HTTP+JSON', default_input_modes: 'list[str]' = <factory>, default_output_modes: 'list[str]' = <factory>, streaming: 'bool' = True, provider: 'dict[str, str] | None' = None, documentation_url: 'str | None' = None, security_schemes: 'dict[str, JsonValue]' = <factory>, security: 'list[dict[str, list[str]]]' = <factory>) -> NoneA2AAgentCard(name: 'str', description: 'str', url: 'str', version: 'str', skills: 'list[A2AAgentSkill]', protocol_version: 'str' = '1.0', preferred_transport: 'str' = 'HTTP+JSON', default_input_modes: 'list[str]' = <factory>, default_output_modes: 'list[str]' = <factory>, streaming: 'bool' = True, provider: 'dict[str, str] | None' = None, documentation_url: 'str | None' = None, security_schemes: 'dict[str, JsonValue]' = <factory>, security: 'list[dict[str, list[str]]]' = <factory>)A2AAgentExecutor
Beta · protocol
from zhivex_ai import A2AAgentExecutorA2AAgentExecutor(agent: 'Agent', *, run_options_resolver: 'ProtocolRunOptionsResolver | None' = None, error_mapper: 'ProtocolErrorMapper | None' = None, on_protocol_event: 'ProtocolEventCallback | None' = None, limits: 'ProtocolLimits | None' = None) -> 'None'In-process A2A v1 executor backed by a configured Zhivex Agent.A2AAgentSkill
Beta · protocol
from zhivex_ai import A2AAgentSkillA2AAgentSkill(id: 'str', name: 'str', description: 'str', tags: 'list[str]' = <factory>, examples: 'list[str]' = <factory>, input_modes: 'list[str] | None' = None, output_modes: 'list[str] | None' = None) -> NoneA2AAgentSkill(id: 'str', name: 'str', description: 'str', tags: 'list[str]' = <factory>, examples: 'list[str]' = <factory>, input_modes: 'list[str] | None' = None, output_modes: 'list[str] | None' = None)A2A_PROTOCOL_VERSION
Beta · protocol
from zhivex_ai import A2A_PROTOCOL_VERSIONA2A_PROTOCOL_VERSION (public type alias or constant; no callable signature)AGENT_EVALUATION_ARTIFACT_SCHEMA_VERSION
Beta · agent
from zhivex_ai import AGENT_EVALUATION_ARTIFACT_SCHEMA_VERSIONAGENT_EVALUATION_ARTIFACT_SCHEMA_VERSION (public type alias or constant; no callable signature)AGENT_RUN_STATE_SCHEMA_VERSION
Beta · agent
from zhivex_ai import AGENT_RUN_STATE_SCHEMA_VERSIONAGENT_RUN_STATE_SCHEMA_VERSION (public type alias or constant; no callable signature)AGUIEvent
Beta · protocol
from zhivex_ai import AGUIEventAGUIEvent(type: 'str', data: 'dict[str, Any]' = <factory>) -> NoneAGUIEvent(type: 'str', data: 'dict[str, Any]' = <factory>)Agent
Stable · agent
from zhivex_ai import AgentAgent(name: 'str', model: 'LanguageModel | RealtimeModel', instructions: 'str | Callable[[AgentContext[AgentDepsT]], str | None | Awaitable[str | None]] | Callable[[AgentContext[AgentDepsT], Agent[AgentDepsT, AgentOutputT]], str | None | Awaitable[str | None]] | None' = None, tools: 'ToolSet | ToolRegistry' = <factory>, skills: 'SkillSet | SkillRegistry' = <factory>, subagents: "dict[str, 'Agent[AgentDepsT, Any]']" = <factory>, memory: 'AgentMemory | None' = None, checkpoint_store: 'AgentCheckpointStore | None' = None, run_store: 'AgentRunStore | None' = None, approval_policy: 'ApprovalPolicy | None' = None, input_guardrails: 'list[InputGuardrail]' = <factory>, output_guardrails: 'list[OutputGuardrail]' = <factory>, tool_execution: 'ToolExecutionOptions | None' = None, run_limits: 'RunLimits' = <factory>, metadata: 'dict[str, Any]' = <factory>, output_type: 'type[AgentOutputT] | None' = None, output_mode: "Literal['auto', 'native', 'prompted']" = 'auto', output_name: 'str | None' = None, output_description: 'str | None' = None, hooks: 'list[AgentHooks]' = <factory>, middleware: 'list[AgentMiddleware]' = <factory>) -> NoneAgent(name: 'str', model: 'LanguageModel | RealtimeModel', instructions: 'str | Callable[[AgentContext[AgentDepsT]], str | None | Awaitable[str | None]] | Callable[[AgentContext[AgentDepsT], Agent[AgentDepsT, AgentOutputT]], str | None | Awaitable[str | None]] | None' = None, tools: 'ToolSet | ToolRegistry' = <factory>, skills: 'SkillSet | SkillRegistry' = <factory>, subagents: "dict[str, 'Agent[AgentDepsT, Any]']" = <factory>, memory: 'AgentMemory | None' = None, checkpoint_store: 'AgentCheckpointStore | None' = None, run_store: 'AgentRunStore | None' = None, approval_policy: 'ApprovalPolicy | None' = None, input_guardrails: 'list[InputGuardrail]' = <factory>, output_guardrails: 'list[OutputGuardrail]' = <factory>, tool_execution: 'ToolExecutionOptions | None' = None, run_limits: 'RunLimits' = <factory>, metadata: 'dict[str, Any]' = <factory>, output_type: 'type[AgentOutputT] | None' = None, output_mode: "Literal['auto', 'native', 'prompted']" = 'auto', output_name: 'str | None' = None, output_description: 'str | None' = None, hooks: 'list[AgentHooks]' = <factory>, middleware: 'list[AgentMiddleware]' = <factory>)AgentCancellationToken
Beta · agent
from zhivex_ai import AgentCancellationTokenAgentCancellationToken(reason: 'str | None' = None) -> NoneIn-process cooperative cancellation signal for one agent run tree.AgentCapabilities
Stable · types
from zhivex_ai import AgentCapabilitiesAgentCapabilities(support_tier: 'AgentSupportTier' = 'tier-c', tool_choice_none: 'bool' = False, approval_requests: 'bool' = False, hosted_web_search: 'bool' = False, hosted_file_search: 'bool' = False, remote_mcp: 'bool' = False, computer_use: 'bool' = False, code_execution: 'bool' = False, toolsets: 'bool' = False) -> NoneAgentCapabilities(support_tier: 'AgentSupportTier' = 'tier-c', tool_choice_none: 'bool' = False, approval_requests: 'bool' = False, hosted_web_search: 'bool' = False, hosted_file_search: 'bool' = False, remote_mcp: 'bool' = False, computer_use: 'bool' = False, code_execution: 'bool' = False, toolsets: 'bool' = False)AgentCheckpoint
Stable · agent
from zhivex_ai import AgentCheckpointAgentCheckpoint(run_id: 'str', session_id: 'str', agent_name: 'str', step_index: 'int', request: 'ModelGenerateInput', response: 'Any', saved_at_ms: 'int', is_final: 'bool' = False) -> NoneAgentCheckpoint(run_id: 'str', session_id: 'str', agent_name: 'str', step_index: 'int', request: 'ModelGenerateInput', response: 'Any', saved_at_ms: 'int', is_final: 'bool' = False)AgentCheckpointEvent
Beta · agent
from zhivex_ai import AgentCheckpointEventAgentCheckpointEvent(type: 'str' = 'checkpoint', checkpoint: 'AgentCheckpoint | None' = None) -> NoneAgentCheckpointEvent(type: 'str' = 'checkpoint', checkpoint: 'AgentCheckpoint | None' = None)AgentChildRun
Stable · agent
from zhivex_ai import AgentChildRunAgentChildRun(run_id: 'str', agent_name: 'str', parent_run_id: 'str', status: 'AgentRunStatus', output_text: 'str' = '', tool_name: 'str | None' = None, error: 'str | None' = None, steps: 'int' = 0, tool_calls: 'int' = 0, tool_errors: 'int' = 0, usage: 'TokenUsage | None' = None) -> NoneAgentChildRun(run_id: 'str', agent_name: 'str', parent_run_id: 'str', status: 'AgentRunStatus', output_text: 'str' = '', tool_name: 'str | None' = None, error: 'str | None' = None, steps: 'int' = 0, tool_calls: 'int' = 0, tool_errors: 'int' = 0, usage: 'TokenUsage | None' = None)AgentContext
Stable · agent
from zhivex_ai import AgentContextAgentContext(run_id: 'str', session_id: 'str', agent_name: 'str', memory_summary: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>, handoff_path: 'list[str]' = <factory>, deps: 'AgentDepsT | None' = None, session: 'AgentSession | None' = None, cancellation_token: 'AgentCancellationToken | None' = None) -> NoneAgentContext(run_id: 'str', session_id: 'str', agent_name: 'str', memory_summary: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>, handoff_path: 'list[str]' = <factory>, deps: 'AgentDepsT | None' = None, session: 'AgentSession | None' = None, cancellation_token: 'AgentCancellationToken | None' = None)AgentDelegationFinishEvent
Beta · agent
from zhivex_ai import AgentDelegationFinishEventAgentDelegationFinishEvent(type: 'str' = 'delegation-finish', agent_name: 'str' = '', handoff_depth: 'int' = 0, finish_reason: 'FinishReason | None' = None) -> NoneAgentDelegationFinishEvent(type: 'str' = 'delegation-finish', agent_name: 'str' = '', handoff_depth: 'int' = 0, finish_reason: 'FinishReason | None' = None)AgentDelegationStartEvent
Beta · agent
from zhivex_ai import AgentDelegationStartEventAgentDelegationStartEvent(type: 'str' = 'delegation-start', agent_name: 'str' = '', handoff_depth: 'int' = 0) -> NoneAgentDelegationStartEvent(type: 'str' = 'delegation-start', agent_name: 'str' = '', handoff_depth: 'int' = 0)AgentErrorEvent
Beta · agent
from zhivex_ai import AgentErrorEventAgentErrorEvent(type: 'str' = 'error', error: 'Exception | None' = None) -> NoneAgentErrorEvent(type: 'str' = 'error', error: 'Exception | None' = None)AgentEvaluationAgentFactory
Beta · agent
from zhivex_ai import AgentEvaluationAgentFactoryAgentEvaluationAgentFactory(*args, **kwargs)AgentEvaluationCase
Beta · agent
from zhivex_ai import AgentEvaluationCaseAgentEvaluationCase(name: 'str', prompt: 'str | None' = None, expectations: 'AgentEvaluationExpectations | None' = None, metadata: 'dict[str, JsonValue]' = <factory>) -> NoneAgentEvaluationCase(name: 'str', prompt: 'str | None' = None, expectations: 'AgentEvaluationExpectations | None' = None, metadata: 'dict[str, JsonValue]' = <factory>)AgentEvaluationCaseResult
Beta · agent
from zhivex_ai import AgentEvaluationCaseResultAgentEvaluationCaseResult(name: 'str', ok: 'bool', output: 'AgentRunResult | None' = None, failures: 'list[str]' = <factory>, metadata: 'dict[str, JsonValue]' = <factory>, trials: 'list[AgentEvaluationTrialResult]' = <factory>) -> NoneAgentEvaluationCaseResult(name: 'str', ok: 'bool', output: 'AgentRunResult | None' = None, failures: 'list[str]' = <factory>, metadata: 'dict[str, JsonValue]' = <factory>, trials: 'list[AgentEvaluationTrialResult]' = <factory>)AgentEvaluationCostEstimator
Beta · agent
from zhivex_ai import AgentEvaluationCostEstimatorAgentEvaluationCostEstimator(*args, **kwargs)AgentEvaluationExpectations
Beta · agent
from zhivex_ai import AgentEvaluationExpectationsAgentEvaluationExpectations(status: 'str | None' = None, output_contains: 'str | None' = None, output_equals: 'str | None' = None, tool_calls: 'list[str]' = <factory>, child_run_count: 'int | None' = None, child_agents: 'list[str]' = <factory>, child_statuses: 'list[str]' = <factory>, finish_reason: 'FinishReason | None' = None, error_contains: 'str | None' = None, workflow_steps: 'list[str]' = <factory>, state_contains: 'list[str]' = <factory>, state_equals: 'dict[str, JsonValue]' = <factory>, failed_steps: 'list[str]' = <factory>) -> NoneAgentEvaluationExpectations(status: 'str | None' = None, output_contains: 'str | None' = None, output_equals: 'str | None' = None, tool_calls: 'list[str]' = <factory>, child_run_count: 'int | None' = None, child_agents: 'list[str]' = <factory>, child_statuses: 'list[str]' = <factory>, finish_reason: 'FinishReason | None' = None, error_contains: 'str | None' = None, workflow_steps: 'list[str]' = <factory>, state_contains: 'list[str]' = <factory>, state_equals: 'dict[str, JsonValue]' = <factory>, failed_steps: 'list[str]' = <factory>)AgentEvaluationExperimentResult
Beta · agent
from zhivex_ai import AgentEvaluationExperimentResultAgentEvaluationExperimentResult(ok: 'bool', baseline: 'str', variants: 'list[AgentEvaluationVariantResult]', gates: 'list[AgentEvaluationGateResult]', metadata: 'dict[str, JsonValue]' = <factory>, schema_version: 'int' = 1) -> NoneAgentEvaluationExperimentResult(ok: 'bool', baseline: 'str', variants: 'list[AgentEvaluationVariantResult]', gates: 'list[AgentEvaluationGateResult]', metadata: 'dict[str, JsonValue]' = <factory>, schema_version: 'int' = 1)AgentEvaluationFixture
Beta · agent
from zhivex_ai import AgentEvaluationFixtureAgentEvaluationFixture(name: 'str', dataset: 'list[AgentEvaluationCase]', expected_ok: 'bool' = True, metadata: 'dict[str, JsonValue]' = <factory>) -> NoneAgentEvaluationFixture(name: 'str', dataset: 'list[AgentEvaluationCase]', expected_ok: 'bool' = True, metadata: 'dict[str, JsonValue]' = <factory>)AgentEvaluationGate
Beta · agent
from zhivex_ai import AgentEvaluationGateAgentEvaluationGate(metric: 'str' = 'pass_rate', minimum: 'float | None' = None, maximum: 'float | None' = None, max_regression: 'float | None' = 0.0) -> NoneAgentEvaluationGate(metric: 'str' = 'pass_rate', minimum: 'float | None' = None, maximum: 'float | None' = None, max_regression: 'float | None' = 0.0)AgentEvaluationGateResult
Beta · agent
from zhivex_ai import AgentEvaluationGateResultAgentEvaluationGateResult(variant: 'str', metric: 'str', value: 'float', baseline_value: 'float', delta: 'float', regression: 'float', ok: 'bool', minimum: 'float | None' = None, maximum: 'float | None' = None, max_regression: 'float | None' = None, failures: 'list[str]' = <factory>) -> NoneAgentEvaluationGateResult(variant: 'str', metric: 'str', value: 'float', baseline_value: 'float', delta: 'float', regression: 'float', ok: 'bool', minimum: 'float | None' = None, maximum: 'float | None' = None, max_regression: 'float | None' = None, failures: 'list[str]' = <factory>)AgentEvaluationJudgeResult
Beta · agent
from zhivex_ai import AgentEvaluationJudgeResultAgentEvaluationJudgeResult(score: 'float', feedback: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>) -> NoneAgentEvaluationJudgeResult(score: 'float', feedback: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>)AgentEvaluationMetric
Beta · agent
from zhivex_ai import AgentEvaluationMetricAgentEvaluationMetric(name: 'str', scorer: 'AgentEvaluationScorer', higher_is_better: 'bool' = True) -> NoneAgentEvaluationMetric(name: 'str', scorer: 'AgentEvaluationScorer', higher_is_better: 'bool' = True)AgentEvaluationReport
Beta · agent
from zhivex_ai import AgentEvaluationReportAgentEvaluationReport(ok: 'bool', total: 'int', passed: 'int', failed: 'int', pass_rate: 'float', failures: 'list[dict[str, JsonValue]]', cases: 'list[dict[str, JsonValue]]', metadata: 'dict[str, JsonValue]' = <factory>, trial_total: 'int' = 0, trial_passed: 'int' = 0, trial_failed: 'int' = 0, trial_pass_rate: 'float' = 1.0, metrics: 'dict[str, float]' = <factory>, gate_failures: 'list[str]' = <factory>, schema_version: 'int' = 1) -> NoneAgentEvaluationReport(ok: 'bool', total: 'int', passed: 'int', failed: 'int', pass_rate: 'float', failures: 'list[dict[str, JsonValue]]', cases: 'list[dict[str, JsonValue]]', metadata: 'dict[str, JsonValue]' = <factory>, trial_total: 'int' = 0, trial_passed: 'int' = 0, trial_failed: 'int' = 0, trial_pass_rate: 'float' = 1.0, metrics: 'dict[str, float]' = <factory>, gate_failures: 'list[str]' = <factory>, schema_version: 'int' = 1)AgentEvaluationResult
Beta · agent
from zhivex_ai import AgentEvaluationResultAgentEvaluationResult(ok: 'bool', cases: 'list[AgentEvaluationCaseResult]') -> NoneAgentEvaluationResult(ok: 'bool', cases: 'list[AgentEvaluationCaseResult]')AgentEvaluationScorer
Beta · agent
from zhivex_ai import AgentEvaluationScorerAgentEvaluationScorer(*args, **kwargs)AgentEvaluationTraceExpectationsExtractor
Beta · agent
from zhivex_ai import AgentEvaluationTraceExpectationsExtractorAgentEvaluationTraceExpectationsExtractor(*args, **kwargs)AgentEvaluationTraceMetadataExtractor
Beta · agent
from zhivex_ai import AgentEvaluationTraceMetadataExtractorAgentEvaluationTraceMetadataExtractor(*args, **kwargs)AgentEvaluationTraceNameExtractor
Beta · agent
from zhivex_ai import AgentEvaluationTraceNameExtractorAgentEvaluationTraceNameExtractor(*args, **kwargs)AgentEvaluationTracePromptExtractor
Beta · agent
from zhivex_ai import AgentEvaluationTracePromptExtractorAgentEvaluationTracePromptExtractor(*args, **kwargs)AgentEvaluationTrajectory
Beta · agent
from zhivex_ai import AgentEvaluationTrajectoryAgentEvaluationTrajectory(run_id: 'str', orchestration_path: 'list[str]', events: 'list[AgentEvaluationTrajectoryEvent]') -> NoneAgentEvaluationTrajectory(run_id: 'str', orchestration_path: 'list[str]', events: 'list[AgentEvaluationTrajectoryEvent]')AgentEvaluationTrajectoryEvent
Beta · agent
from zhivex_ai import AgentEvaluationTrajectoryEventAgentEvaluationTrajectoryEvent(type: 'str', agent_name: 'str | None' = None, name: 'str | None' = None, source_agent: 'str | None' = None, target_agent: 'str | None' = None, status: 'str | None' = None) -> NoneA deliberately redacted projection of one runtime trace event.AgentEvaluationTrialResult
Beta · agent
from zhivex_ai import AgentEvaluationTrialResultAgentEvaluationTrialResult(repetition: 'int', ok: 'bool', output: 'AgentRunResult | None' = None, failures: 'list[str]' = <factory>, latency_ms: 'float' = 0.0, usage: 'TokenUsage | None' = None, cost: 'float | None' = None, trajectory: 'AgentEvaluationTrajectory | None' = None) -> NoneAgentEvaluationTrialResult(repetition: 'int', ok: 'bool', output: 'AgentRunResult | None' = None, failures: 'list[str]' = <factory>, latency_ms: 'float' = 0.0, usage: 'TokenUsage | None' = None, cost: 'float | None' = None, trajectory: 'AgentEvaluationTrajectory | None' = None)AgentEvaluationVariant
Beta · agent
from zhivex_ai import AgentEvaluationVariantAgentEvaluationVariant(name: 'str', agent: 'Agent | AgentEvaluationAgentFactory', metadata: 'dict[str, JsonValue]' = <factory>) -> NoneAgentEvaluationVariant(name: 'str', agent: 'Agent | AgentEvaluationAgentFactory', metadata: 'dict[str, JsonValue]' = <factory>)AgentEvaluationVariantResult
Beta · agent
from zhivex_ai import AgentEvaluationVariantResultAgentEvaluationVariantResult(name: 'str', result: 'AgentEvaluationResult', report: 'AgentEvaluationReport', metrics: 'dict[str, float]', metadata: 'dict[str, JsonValue]' = <factory>) -> NoneAgentEvaluationVariantResult(name: 'str', result: 'AgentEvaluationResult', report: 'AgentEvaluationReport', metrics: 'dict[str, float]', metadata: 'dict[str, JsonValue]' = <factory>)AgentEvent
Beta · agent
from zhivex_ai import AgentEventAgentEvent (public type alias or constant; no callable signature)AgentEventDeliveryError
Stable · errors
from zhivex_ai import AgentEventDeliveryErrorAgentEventDeliveryError(run_id: 'str', *, event_type: 'str', durable_state_committed: 'bool') -> 'None'Raised when an application event callback fails during an agent run.AgentFinishEvent
Beta · agent
from zhivex_ai import AgentFinishEventAgentFinishEvent(type: 'str' = 'finish', run_id: 'str' = '', session_id: 'str' = '', text: 'str' = '', finish_reason: 'FinishReason | None' = None) -> NoneAgentFinishEvent(type: 'str' = 'finish', run_id: 'str' = '', session_id: 'str' = '', text: 'str' = '', finish_reason: 'FinishReason | None' = None)AgentGroupMember
Beta · agent
from zhivex_ai import AgentGroupMemberAgentGroupMember(name: 'str', agent: 'Agent', prompt: 'str | None' = None, session: 'AgentSession | None' = None, idempotency_key: 'str | None' = None) -> NoneAgentGroupMember(name: 'str', agent: 'Agent', prompt: 'str | None' = None, session: 'AgentSession | None' = None, idempotency_key: 'str | None' = None)AgentGroupMemberResult
Beta · agent
from zhivex_ai import AgentGroupMemberResultAgentGroupMemberResult(name: 'str', output: 'AgentRunResult | None' = None, error: 'Exception | None' = None) -> NoneAgentGroupMemberResult(name: 'str', output: 'AgentRunResult | None' = None, error: 'Exception | None' = None)AgentGroupRunResult
Beta · agent
from zhivex_ai import AgentGroupRunResultAgentGroupRunResult(parent_run_id: 'str | None', outputs: 'list[AgentGroupMemberResult]') -> NoneAgentGroupRunResult(parent_run_id: 'str | None', outputs: 'list[AgentGroupMemberResult]')AgentGuardrailEvent
Beta · agent
from zhivex_ai import AgentGuardrailEventAgentGuardrailEvent(type: 'str' = 'guardrail', stage: "Literal['input', 'output']" = 'input', guardrail_name: 'str' = '', triggered: 'bool' = False, reason: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneAgentGuardrailEvent(type: 'str' = 'guardrail', stage: "Literal['input', 'output']" = 'input', guardrail_name: 'str' = '', triggered: 'bool' = False, reason: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)AgentHandoff
Stable · agent
from zhivex_ai import AgentHandoffAgentHandoff(target_agent: 'str', input: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneAgentHandoff(target_agent: 'str', input: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)AgentHandoffEvent
Beta · agent
from zhivex_ai import AgentHandoffEventAgentHandoffEvent(type: 'str' = 'handoff', handoff: 'AgentHandoff | None' = None) -> NoneAgentHandoffEvent(type: 'str' = 'handoff', handoff: 'AgentHandoff | None' = None)AgentHandoffFailedEvent
Beta · agent
from zhivex_ai import AgentHandoffFailedEventAgentHandoffFailedEvent(type: 'str' = 'handoff-failed', source_agent: 'str' = '', target_agent: 'str' = '', reason: 'str | None' = None) -> NoneAgentHandoffFailedEvent(type: 'str' = 'handoff-failed', source_agent: 'str' = '', target_agent: 'str' = '', reason: 'str | None' = None)AgentHandoffRequestedEvent
Beta · agent
from zhivex_ai import AgentHandoffRequestedEventAgentHandoffRequestedEvent(type: 'str' = 'handoff-requested', handoff: 'AgentHandoff | None' = None) -> NoneAgentHandoffRequestedEvent(type: 'str' = 'handoff-requested', handoff: 'AgentHandoff | None' = None)AgentHandoffResolvedEvent
Beta · agent
from zhivex_ai import AgentHandoffResolvedEventAgentHandoffResolvedEvent(type: 'str' = 'handoff-resolved', source_agent: 'str' = '', target_agent: 'str' = '') -> NoneAgentHandoffResolvedEvent(type: 'str' = 'handoff-resolved', source_agent: 'str' = '', target_agent: 'str' = '')AgentHooks
Stable · agent
from zhivex_ai import AgentHooksAgentHooks()No-op lifecycle hooks that applications can override selectively.AgentLiveEvent
Experimental · agent
from zhivex_ai import AgentLiveEventAgentLiveEvent (public type alias or constant; no callable signature)AgentMemoryState
Stable · agent
from zhivex_ai import AgentMemoryStateAgentMemoryState(messages: 'list[ModelMessage]' = <factory>, summary: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneAgentMemoryState(messages: 'list[ModelMessage]' = <factory>, summary: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)AgentMiddleware
Stable · agent
from zhivex_ai import AgentMiddlewareAgentMiddleware(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...AgentMiddlewareNext
Stable · agent
from zhivex_ai import AgentMiddlewareNextAgentMiddlewareNext(*args, **kwargs)AgentObserver
Stable · agent
from zhivex_ai import AgentObserverAgentObserver(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...AgentRegistry
Stable · agent
from zhivex_ai import AgentRegistryAgentRegistry(agents: 'dict[str, Agent] | None' = None) -> 'None'AgentReplayEvent
Stable · agent
from zhivex_ai import AgentReplayEventAgentReplayEvent(type: 'str', run_id: 'str', step: 'int | None' = None, status: 'str | None' = None, name: 'str | None' = None, data: 'JsonValue | None' = None) -> NoneAgentReplayEvent(type: 'str', run_id: 'str', step: 'int | None' = None, status: 'str | None' = None, name: 'str | None' = None, data: 'JsonValue | None' = None)AgentReplayResult
Stable · agent
from zhivex_ai import AgentReplayResultAgentReplayResult(snapshot: 'AgentRunSnapshot', timeline: 'list[AgentReplayEvent]') -> NoneAgentReplayResult(snapshot: 'AgentRunSnapshot', timeline: 'list[AgentReplayEvent]')AgentResolver
Beta · protocol
from zhivex_ai import AgentResolverAgentResolver (public type alias or constant; no callable signature)AgentRunCancelled
Stable · errors
from zhivex_ai import AgentRunCancelledAgentRunCancelled(run_id: 'str', *, reason: 'str | None' = None) -> 'None'Raised when an atomic cancellation wins against an active agent worker.AgentRunRequest
Stable · agent
from zhivex_ai import AgentRunRequestAgentRunRequest(agent: 'Agent[AgentDepsT, AgentOutputT]', session: 'AgentSession | None' = None, prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, deps: 'AgentDepsT | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneMutable request passed through agent run middleware.AgentRunResult
Stable · agent
from zhivex_ai import AgentRunResultAgentRunResult(run_id: 'str', agent_name: 'str', session: 'AgentSession', text: 'str', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, steps: 'list[GenerateTextStep]' = <factory>, messages: 'list[ModelMessage]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, artifacts: 'list[SkillArtifact]' = <factory>, trace: 'AgentTrace | None' = None, handoff: 'AgentHandoff | None' = None, orchestration_path: 'list[str]' = <factory>, resumed_from_checkpoint: 'AgentCheckpoint | None' = None, state: 'AgentRunState | None' = None, output: 'AgentOutputT | None' = None) -> NoneAgentRunResult(run_id: 'str', agent_name: 'str', session: 'AgentSession', text: 'str', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, steps: 'list[GenerateTextStep]' = <factory>, messages: 'list[ModelMessage]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, artifacts: 'list[SkillArtifact]' = <factory>, trace: 'AgentTrace | None' = None, handoff: 'AgentHandoff | None' = None, orchestration_path: 'list[str]' = <factory>, resumed_from_checkpoint: 'AgentCheckpoint | None' = None, state: 'AgentRunState | None' = None, output: 'AgentOutputT | None' = None)AgentRunSnapshot
Stable · agent
from zhivex_ai import AgentRunSnapshotAgentRunSnapshot(run_id: 'str', agent_name: 'str', status: 'str', provider: 'str', model_id: 'str', steps: 'int', tool_calls: 'int', child_runs: 'int', output_text: 'str', error: 'str | None' = None) -> NoneAgentRunSnapshot(run_id: 'str', agent_name: 'str', status: 'str', provider: 'str', model_id: 'str', steps: 'int', tool_calls: 'int', child_runs: 'int', output_text: 'str', error: 'str | None' = None)AgentRunStartEvent
Beta · agent
from zhivex_ai import AgentRunStartEventAgentRunStartEvent(type: 'str' = 'run-start', run_id: 'str' = '', session_id: 'str' = '', agent_name: 'str' = '') -> NoneAgentRunStartEvent(type: 'str' = 'run-start', run_id: 'str' = '', session_id: 'str' = '', agent_name: 'str' = '')AgentRunState
Stable · agent
from zhivex_ai import AgentRunStateAgentRunState(run_id: 'str', agent_name: 'str', provider: 'str', model_id: 'str', schema_version: 'int' = 1, revision: 'int' = 0, status: 'AgentRunStatus' = 'running', session_id: 'str | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, started_at_ms: 'int | None' = None, updated_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None, current_step: 'int' = 0, steps: 'list[AgentRunStep]' = <factory>, child_runs: 'list[AgentChildRun]' = <factory>, pending_approvals: 'list[PendingApproval]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, usage: 'TokenUsage | None' = None, output_text: 'str' = '', finish_reason: 'FinishReason | None' = None, error: 'str | None' = None, cancellation_reason: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>) -> NoneAgentRunState(run_id: 'str', agent_name: 'str', provider: 'str', model_id: 'str', schema_version: 'int' = 1, revision: 'int' = 0, status: 'AgentRunStatus' = 'running', session_id: 'str | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, started_at_ms: 'int | None' = None, updated_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None, current_step: 'int' = 0, steps: 'list[AgentRunStep]' = <factory>, child_runs: 'list[AgentChildRun]' = <factory>, pending_approvals: 'list[PendingApproval]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, usage: 'TokenUsage | None' = None, output_text: 'str' = '', finish_reason: 'FinishReason | None' = None, error: 'str | None' = None, cancellation_reason: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>)AgentRunStatus
Stable · agent
from zhivex_ai import AgentRunStatusAgentRunStatus(*args, **kwargs)AgentRunStep
Stable · agent
from zhivex_ai import AgentRunStepAgentRunStep(index: 'int', status: 'AgentRunStatus', tool_calls: 'list[ToolCall]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, usage: 'TokenUsage | None' = None, messages: 'list[ModelMessage]' = <factory>, error: 'str | None' = None, started_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None) -> NoneAgentRunStep(index: 'int', status: 'AgentRunStatus', tool_calls: 'list[ToolCall]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, usage: 'TokenUsage | None' = None, messages: 'list[ModelMessage]' = <factory>, error: 'str | None' = None, started_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None)AgentRunStore
Stable · agent
from zhivex_ai import AgentRunStoreAgentRunStore(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...AgentRunTreeCancellationResult
Stable · agent
from zhivex_ai import AgentRunTreeCancellationResultAgentRunTreeCancellationResult(root: 'AgentRunState | None', cancelled: 'list[AgentRunState]' = <factory>, missing_parent_lookup: 'bool' = False) -> NoneAgentRunTreeCancellationResult(root: 'AgentRunState | None', cancelled: 'list[AgentRunState]' = <factory>, missing_parent_lookup: 'bool' = False)AgentRunTreeNode
Beta · observability
from zhivex_ai import AgentRunTreeNodeAgentRunTreeNode(run_id: 'str', agent_name: 'str', parent_run_id: 'str | None', status: 'str', snapshot: 'AgentRunSnapshot', children: "list['AgentRunTreeNode']") -> NoneAgentRunTreeNode(run_id: 'str', agent_name: 'str', parent_run_id: 'str | None', status: 'str', snapshot: 'AgentRunSnapshot', children: "list['AgentRunTreeNode']")AgentRunTreeSnapshot
Beta · observability
from zhivex_ai import AgentRunTreeSnapshotAgentRunTreeSnapshot(root: 'AgentRunTreeNode', total_runs: 'int') -> NoneAgentRunTreeSnapshot(root: 'AgentRunTreeNode', total_runs: 'int')AgentRuntime
Stable · agent
from zhivex_ai import AgentRuntimeAgentRuntime(*, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None) -> 'None'AgentSession
Stable · agent
from zhivex_ai import AgentSessionAgentSession(id: 'str' = <factory>, messages: 'list[ModelMessage]' = <factory>, summary: 'str | None' = None, state: 'dict[str, JsonValue]' = <factory>, metadata: 'dict[str, Any]' = <factory>) -> NoneAgentSession(id: 'str' = <factory>, messages: 'list[ModelMessage]' = <factory>, summary: 'str | None' = None, state: 'dict[str, JsonValue]' = <factory>, metadata: 'dict[str, Any]' = <factory>)AgentSkillActivatedEvent
Stable · agent
from zhivex_ai import AgentSkillActivatedEventAgentSkillActivatedEvent(type: 'str' = 'skill-activated', skill_name: 'str' = '', activation: 'SkillActivationMode' = 'explicit', path: 'str | None' = None, description: 'str | None' = None) -> NoneAgentSkillActivatedEvent(type: 'str' = 'skill-activated', skill_name: 'str' = '', activation: 'SkillActivationMode' = 'explicit', path: 'str | None' = None, description: 'str | None' = None)AgentSkillArtifactCreatedEvent
Beta · agent
from zhivex_ai import AgentSkillArtifactCreatedEventAgentSkillArtifactCreatedEvent(type: 'str' = 'skill-artifact-created', skill_name: 'str' = '', entrypoint: 'str' = '', artifact: 'SkillArtifact | None' = None) -> NoneAgentSkillArtifactCreatedEvent(type: 'str' = 'skill-artifact-created', skill_name: 'str' = '', entrypoint: 'str' = '', artifact: 'SkillArtifact | None' = None)AgentSkillDependencyCheckEvent
Beta · agent
from zhivex_ai import AgentSkillDependencyCheckEventAgentSkillDependencyCheckEvent(type: 'str' = 'skill-dependency-check', skill_name: 'str' = '', dependency_type: 'str' = '', dependency_value: 'str' = '', available: 'bool' = True) -> NoneAgentSkillDependencyCheckEvent(type: 'str' = 'skill-dependency-check', skill_name: 'str' = '', dependency_type: 'str' = '', dependency_value: 'str' = '', available: 'bool' = True)AgentSkillExecutionFinishEvent
Beta · agent
from zhivex_ai import AgentSkillExecutionFinishEventAgentSkillExecutionFinishEvent(type: 'str' = 'skill-execution-finish', skill_name: 'str' = '', entrypoint: 'str' = '', ok: 'bool' = True) -> NoneAgentSkillExecutionFinishEvent(type: 'str' = 'skill-execution-finish', skill_name: 'str' = '', entrypoint: 'str' = '', ok: 'bool' = True)AgentSkillExecutionStartEvent
Beta · agent
from zhivex_ai import AgentSkillExecutionStartEventAgentSkillExecutionStartEvent(type: 'str' = 'skill-execution-start', skill_name: 'str' = '', entrypoint: 'str' = '') -> NoneAgentSkillExecutionStartEvent(type: 'str' = 'skill-execution-start', skill_name: 'str' = '', entrypoint: 'str' = '')AgentSkillResolvedEvent
Beta · agent
from zhivex_ai import AgentSkillResolvedEventAgentSkillResolvedEvent(type: 'str' = 'skill-resolved', skill_name: 'str' = '', skill_version: 'str | None' = None, entrypoints: 'list[str]' = <factory>) -> NoneAgentSkillResolvedEvent(type: 'str' = 'skill-resolved', skill_name: 'str' = '', skill_version: 'str | None' = None, entrypoints: 'list[str]' = <factory>)AgentSkillSkippedEvent
Stable · agent
from zhivex_ai import AgentSkillSkippedEventAgentSkillSkippedEvent(type: 'str' = 'skill-skipped', skill_name: 'str' = '', activation: 'SkillActivationMode' = 'sticky', reason: 'str' = '', path: 'str | None' = None) -> NoneAgentSkillSkippedEvent(type: 'str' = 'skill-skipped', skill_name: 'str' = '', activation: 'SkillActivationMode' = 'sticky', reason: 'str' = '', path: 'str | None' = None)AgentStreamResult
Stable · agent
from zhivex_ai import AgentStreamResultAgentStreamResult(runner: 'asyncio.Task[AgentRunResult[AgentOutputT]]', broadcast: 'Broadcast[AgentEvent]') -> 'None'Abstract base class for generic types.
On Python 3.12 and newer, generic classes implicitly inherit from
Generic when they declare a parameter list after the class's name::
class Mapping[KT, VT]:
def __getitem__(self, key: KT) -> VT:
...
# Etc.
On older versions of Python, however, generic classes have to
explicitly inherit from Generic.
After a class has been declared to be generic, it can then be used as
follows::
def lookup_name[KT, VT](mapping: Mapping[KT, VT], key: KT, default: VT) -> VT:
try:
return mapping[key]
except KeyError:
return defaultAgentSummaryUpdateEvent
Beta · agent
from zhivex_ai import AgentSummaryUpdateEventAgentSummaryUpdateEvent(type: 'str' = 'summary-update', summary: 'str | None' = None) -> NoneAgentSummaryUpdateEvent(type: 'str' = 'summary-update', summary: 'str | None' = None)AgentSupportTier
Stable · types
from zhivex_ai import AgentSupportTierAgentSupportTier(*args, **kwargs)AgentTextDeltaEvent
Beta · agent
from zhivex_ai import AgentTextDeltaEventAgentTextDeltaEvent(type: 'str' = 'text-delta', text_delta: 'str' = '') -> NoneAgentTextDeltaEvent(type: 'str' = 'text-delta', text_delta: 'str' = '')AgentToolApprovalEvent
Stable · agent
from zhivex_ai import AgentToolApprovalEventAgentToolApprovalEvent(type: 'str' = 'tool-approval', tool_name: 'str' = '', tool_input: 'Any' = None, approved: 'bool' = True, reason: 'str | None' = None, provider: 'str | None' = None, provider_managed: 'bool' = False, approval_request_id: 'str | None' = None, tool_source: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneAgentToolApprovalEvent(type: 'str' = 'tool-approval', tool_name: 'str' = '', tool_input: 'Any' = None, approved: 'bool' = True, reason: 'str | None' = None, provider: 'str | None' = None, provider_managed: 'bool' = False, approval_request_id: 'str | None' = None, tool_source: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)AgentToolCallEvent
Beta · agent
from zhivex_ai import AgentToolCallEventAgentToolCallEvent(type: 'str' = 'tool-call', tool_call: 'ToolCall' = <factory>) -> NoneAgentToolCallEvent(type: 'str' = 'tool-call', tool_call: 'ToolCall' = <factory>)AgentToolResultEvent
Beta · agent
from zhivex_ai import AgentToolResultEventAgentToolResultEvent(type: 'str' = 'tool-result', tool_result: 'ToolExecutionResult' = <factory>) -> NoneAgentToolResultEvent(type: 'str' = 'tool-result', tool_result: 'ToolExecutionResult' = <factory>)AgentTrace
Stable · agent
from zhivex_ai import AgentTraceAgentTrace(run_id: 'str', session_id: 'str', agent_name: 'str', started_at_ms: 'int', finished_at_ms: 'int | None' = None, events: 'list[AgentEvent]' = <factory>, orchestration_path: 'list[str]' = <factory>, segments: 'list[AgentTraceSegment]' = <factory>, tool_call_count: 'int' = 0, approval_count: 'int' = 0, guardrail_trigger_count: 'int' = 0, checkpoint_count: 'int' = 0, handoff_count: 'int' = 0) -> NoneAgentTrace(run_id: 'str', session_id: 'str', agent_name: 'str', started_at_ms: 'int', finished_at_ms: 'int | None' = None, events: 'list[AgentEvent]' = <factory>, orchestration_path: 'list[str]' = <factory>, segments: 'list[AgentTraceSegment]' = <factory>, tool_call_count: 'int' = 0, approval_count: 'int' = 0, guardrail_trigger_count: 'int' = 0, checkpoint_count: 'int' = 0, handoff_count: 'int' = 0)AgentTraceArtifact
Beta · observability
from zhivex_ai import AgentTraceArtifactAgentTraceArtifact(run_id: 'str', agent_name: 'str', provider: 'str', model_id: 'str', status: 'str', steps: 'list[AgentTraceStep]', child_runs: 'list[dict[str, Any]]', events: 'list[dict[str, Any]]', usage: 'TokenUsage | None' = None, output_preview: 'str' = '', output_text: 'str | None' = None, error: 'str | None' = None, cancellation_reason: 'str | None' = None, started_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None, duration_ms: 'int | None' = None) -> NoneAgentTraceArtifact(run_id: 'str', agent_name: 'str', provider: 'str', model_id: 'str', status: 'str', steps: 'list[AgentTraceStep]', child_runs: 'list[dict[str, Any]]', events: 'list[dict[str, Any]]', usage: 'TokenUsage | None' = None, output_preview: 'str' = '', output_text: 'str | None' = None, error: 'str | None' = None, cancellation_reason: 'str | None' = None, started_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None, duration_ms: 'int | None' = None)AgentTraceCollector
Beta · observability
from zhivex_ai import AgentTraceCollectorAgentTraceCollector() -> 'None'AgentTraceSegment
Beta · agent
from zhivex_ai import AgentTraceSegmentAgentTraceSegment(agent_name: 'str', started_at_ms: 'int', finished_at_ms: 'int | None' = None) -> NoneAgentTraceSegment(agent_name: 'str', started_at_ms: 'int', finished_at_ms: 'int | None' = None)AgentTraceStep
Beta · observability
from zhivex_ai import AgentTraceStepAgentTraceStep(index: 'int', status: 'str', tool_calls: 'list[dict[str, Any]]', tool_results: 'int', usage: 'TokenUsage | None' = None, error: 'str | None' = None) -> NoneAgentTraceStep(index: 'int', status: 'str', tool_calls: 'list[dict[str, Any]]', tool_results: 'int', usage: 'TokenUsage | None' = None, error: 'str | None' = None)AgentTraceSummary
Beta · observability
from zhivex_ai import AgentTraceSummaryAgentTraceSummary(run_id: 'str', agent_name: 'str', provider: 'str', model_id: 'str', status: 'str', steps: 'int', child_runs: 'int', tool_calls: 'int', tool_errors: 'int', usage: 'TokenUsage | None' = None, cost: 'CostEstimate | None' = None, error: 'str | None' = None, duration_ms: 'int | None' = None) -> NoneAgentTraceSummary(run_id: 'str', agent_name: 'str', provider: 'str', model_id: 'str', status: 'str', steps: 'int', child_runs: 'int', tool_calls: 'int', tool_errors: 'int', usage: 'TokenUsage | None' = None, cost: 'CostEstimate | None' = None, error: 'str | None' = None, duration_ms: 'int | None' = None)AnyToolDefinition
Beta · types
from zhivex_ai import AnyToolDefinitionAnyToolDefinition (public type alias or constant; no callable signature)ApprovalDecision
Stable · agent
from zhivex_ai import ApprovalDecisionApprovalDecision(approved: 'bool', reason: 'str | None' = None, suspend: 'bool' = False, approval_id: 'str | None' = None) -> NoneApprovalDecision(approved: 'bool', reason: 'str | None' = None, suspend: 'bool' = False, approval_id: 'str | None' = None)ApprovalPolicyOptions
Beta · safety
from zhivex_ai import ApprovalPolicyOptionsApprovalPolicyOptions(preset: 'ApprovalPolicyPreset' = 'review_sensitive', sensitive_tool_names: 'list[str]' = <factory>, allow_tool_names: 'list[str]' = <factory>, deny_tool_names: 'list[str]' = <factory>) -> NoneApprovalPolicyOptions(preset: 'ApprovalPolicyPreset' = 'review_sensitive', sensitive_tool_names: 'list[str]' = <factory>, allow_tool_names: 'list[str]' = <factory>, deny_tool_names: 'list[str]' = <factory>)ApprovalPolicyPreset
Beta · safety
from zhivex_ai import ApprovalPolicyPresetApprovalPolicyPreset(*args, **kwargs)AudioFrame
Beta · types
from zhivex_ai import AudioFrameAudioFrame(data: 'bytes | bytearray | memoryview | str', media_type: 'str', sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, timestamp_ms: 'int | None' = None, is_final: 'bool' = False) -> NoneAudioFrame(data: 'bytes | bytearray | memoryview | str', media_type: 'str', sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, timestamp_ms: 'int | None' = None, is_final: 'bool' = False)AudioInput
Beta · types
from zhivex_ai import AudioInputAudioInput(data: 'bytes | bytearray | memoryview | str', media_type: 'str', filename: 'str | None' = None) -> NoneAudioInput(data: 'bytes | bytearray | memoryview | str', media_type: 'str', filename: 'str | None' = None)AzureOpenAIMcpApprovalRequest
Beta · types
from zhivex_ai import AzureOpenAIMcpApprovalRequestAzureOpenAIMcpApprovalRequest(type: "Literal['mcp_approval_request']" = 'mcp_approval_request', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '') -> NoneAzureOpenAIMcpApprovalRequest(type: "Literal['mcp_approval_request']" = 'mcp_approval_request', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '')AzureOpenAIMcpApprovalResponse
Beta · types
from zhivex_ai import AzureOpenAIMcpApprovalResponseAzureOpenAIMcpApprovalResponse(type: "Literal['mcp_approval_response']" = 'mcp_approval_response', approval_request_id: 'str' = '', approve: 'bool' = False, id: 'str | None' = None, reason: 'str | None' = None) -> NoneAzureOpenAIMcpApprovalResponse(type: "Literal['mcp_approval_response']" = 'mcp_approval_response', approval_request_id: 'str' = '', approve: 'bool' = False, id: 'str | None' = None, reason: 'str | None' = None)AzureOpenAIMcpCall
Beta · types
from zhivex_ai import AzureOpenAIMcpCallAzureOpenAIMcpCall(type: "Literal['mcp_call']" = 'mcp_call', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '', approval_request_id: 'str | None' = None, error: 'str | None' = None, output: 'str | None' = None, status: "Literal['in_progress', 'completed', 'incomplete', 'calling', 'failed'] | None" = None) -> NoneAzureOpenAIMcpCall(type: "Literal['mcp_call']" = 'mcp_call', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '', approval_request_id: 'str | None' = None, error: 'str | None' = None, output: 'str | None' = None, status: "Literal['in_progress', 'completed', 'incomplete', 'calling', 'failed'] | None" = None)AzureOpenAIMcpListTools
Beta · types
from zhivex_ai import AzureOpenAIMcpListToolsAzureOpenAIMcpListTools(type: "Literal['mcp_list_tools']" = 'mcp_list_tools', id: 'str | None' = None, server_label: 'str | None' = None, tools: 'JsonValue | None' = None) -> NoneAzureOpenAIMcpListTools(type: "Literal['mcp_list_tools']" = 'mcp_list_tools', id: 'str | None' = None, server_label: 'str | None' = None, tools: 'JsonValue | None' = None)AzureOpenAIProviderData
Beta · types
from zhivex_ai import AzureOpenAIProviderDataAzureOpenAIProviderData (public type alias or constant; no callable signature)AzureOpenAIResponseReference
Beta · types
from zhivex_ai import AzureOpenAIResponseReferenceAzureOpenAIResponseReference(response_id: 'str') -> NoneAzureOpenAIResponseReference(response_id: 'str')BatchesClient
Beta · types
from zhivex_ai import BatchesClientBatchesClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...BudgetGuard
Beta · safety
from zhivex_ai import BudgetGuardBudgetGuard(max_steps: 'int | None' = None, max_tool_calls: 'int | None' = None, max_tool_errors: 'int | None' = None, max_input_tokens: 'int | None' = None, max_output_tokens: 'int | None' = None, max_total_tokens: 'int | None' = None, include_child_runs: 'bool' = True) -> NoneBudgetGuard(max_steps: 'int | None' = None, max_tool_calls: 'int | None' = None, max_tool_errors: 'int | None' = None, max_input_tokens: 'int | None' = None, max_output_tokens: 'int | None' = None, max_total_tokens: 'int | None' = None, include_child_runs: 'bool' = True)CachedContent
Beta · types
from zhivex_ai import CachedContentCachedContent(name: 'str', model: 'str | None' = None, display_name: 'str | None' = None, create_time: 'str | None' = None, update_time: 'str | None' = None, expire_time: 'str | None' = None, usage_metadata: 'dict[str, Any]' = <factory>, metadata: 'dict[str, Any]' = <factory>, raw_response: 'Any' = None) -> NoneCachedContent(name: 'str', model: 'str | None' = None, display_name: 'str | None' = None, create_time: 'str | None' = None, update_time: 'str | None' = None, expire_time: 'str | None' = None, usage_metadata: 'dict[str, Any]' = <factory>, metadata: 'dict[str, Any]' = <factory>, raw_response: 'Any' = None)CachedContentListResult
Beta · types
from zhivex_ai import CachedContentListResultCachedContentListResult(cached_contents: 'list[CachedContent]' = <factory>, next_page_token: 'str | None' = None, raw_response: 'Any' = None) -> NoneCachedContentListResult(cached_contents: 'list[CachedContent]' = <factory>, next_page_token: 'str | None' = None, raw_response: 'Any' = None)CachedContentsClient
Beta · types
from zhivex_ai import CachedContentsClientCachedContentsClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...CallbackWorkflowAdapter
Stable · workflow
from zhivex_ai import CallbackWorkflowAdapterCallbackWorkflowAdapter(backend: 'str', callback: 'WorkflowStepExecutor', capabilities: 'WorkflowAdapterCapabilities' = <factory>) -> NoneCallbackWorkflowAdapter(backend: 'str', callback: 'WorkflowStepExecutor', capabilities: 'WorkflowAdapterCapabilities' = <factory>)CatalogProviderId
Stable · catalog
from zhivex_ai import CatalogProviderIdCatalogProviderId (public type alias or constant; no callable signature)str(object='') -> str
str(bytes_or_buffer[, encoding[, errors]]) -> str
Create a new string object from the given object. If encoding or
errors is specified, then the object must expose a data buffer
that will be decoded using the given encoding and error handler.
Otherwise, returns the result of object.__str__() (if defined)
or repr(object).
encoding defaults to 'utf-8'.
errors defaults to 'strict'.CircuitBreakerState
Beta · middleware
from zhivex_ai import CircuitBreakerStateCircuitBreakerState(failures: 'int' = 0, opened_at: 'int | None' = None) -> NoneCircuitBreakerState(failures: 'int' = 0, opened_at: 'int | None' = None)ConfigurationError
Stable · errors
from zhivex_ai import ConfigurationErrorConfigurationError (public type alias or constant; no callable signature)Common base class for all non-exit exceptions.ContainersClient
Beta · types
from zhivex_ai import ContainersClientContainersClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...ContentPart
Beta · types
from zhivex_ai import ContentPartContentPart (public type alias or constant; no callable signature)ConversationsClient
Beta · types
from zhivex_ai import ConversationsClientConversationsClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...CostEstimate
Beta · observability
from zhivex_ai import CostEstimateCostEstimate(input_cost: 'float | None' = None, output_cost: 'float | None' = None, total_cost: 'float | None' = None, currency: 'str' = 'USD', usage: 'TokenUsage | None' = None) -> NoneCostEstimate(input_cost: 'float | None' = None, output_cost: 'float | None' = None, total_cost: 'float | None' = None, currency: 'str' = 'USD', usage: 'TokenUsage | None' = None)CountTokensClient
Beta · types
from zhivex_ai import CountTokensClientCountTokensClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...CountTokensResult
Beta · types
from zhivex_ai import CountTokensResultCountTokensResult(total_tokens: 'int | None' = None, cached_content_token_count: 'int | None' = None, total_billable_characters: 'int | None' = None, details: 'list[TokenCountDetail]' = <factory>, raw_response: 'Any' = None) -> NoneCountTokensResult(total_tokens: 'int | None' = None, cached_content_token_count: 'int | None' = None, total_billable_characters: 'int | None' = None, details: 'list[TokenCountDetail]' = <factory>, raw_response: 'Any' = None)DynamicInstructions
Stable · agent
from zhivex_ai import DynamicInstructionsDynamicInstructions(*args, **kwargs)EmbedOutput
Stable · types
from zhivex_ai import EmbedOutputEmbedOutput(embeddings: 'list[list[float]]', usage: 'TokenUsage | None' = None, raw_response: 'Any' = None, values: 'list[str]' = <factory>) -> NoneEmbedOutput(embeddings: 'list[list[float]]', usage: 'TokenUsage | None' = None, raw_response: 'Any' = None, values: 'list[str]' = <factory>)EmbeddingContent
Stable · types
from zhivex_ai import EmbeddingContentEmbeddingContent (public type alias or constant; no callable signature)EmbeddingModel
Stable · types
from zhivex_ai import EmbeddingModelEmbeddingModel(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...EmbeddingPart
Beta · types
from zhivex_ai import EmbeddingPartEmbeddingPart (public type alias or constant; no callable signature)FilePart
Beta · types
from zhivex_ai import FilePartFilePart(type: "Literal['file']" = 'file', data: 'str | None' = None, text: 'str | None' = None, document_content: 'list[JsonValue] | None' = None, media_type: 'str | None' = None, filename: 'str | None' = None, file_id: 'str | None' = None, file_uri: 'str | None' = None, url: 'str | None' = None, title: 'str | None' = None, context: 'str | None' = None, citations_enabled: 'bool | None' = None, cache_control: 'dict[str, Any] | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneFilePart(type: "Literal['file']" = 'file', data: 'str | None' = None, text: 'str | None' = None, document_content: 'list[JsonValue] | None' = None, media_type: 'str | None' = None, filename: 'str | None' = None, file_id: 'str | None' = None, file_uri: 'str | None' = None, url: 'str | None' = None, title: 'str | None' = None, context: 'str | None' = None, citations_enabled: 'bool | None' = None, cache_control: 'dict[str, Any] | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)FileSearchBatch
Beta · types
from zhivex_ai import FileSearchBatchFileSearchBatch(name: 'str', file_search_store_name: 'str | None' = None, state: 'str | None' = None, create_time: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>, raw_response: 'Any' = None) -> NoneFileSearchBatch(name: 'str', file_search_store_name: 'str | None' = None, state: 'str | None' = None, create_time: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>, raw_response: 'Any' = None)FileSearchDocument
Beta · types
from zhivex_ai import FileSearchDocumentFileSearchDocument(name: 'str', display_name: 'str | None' = None, custom_metadata: 'list[dict[str, Any]]' = <factory>, state: 'str | None' = None, size_bytes: 'int | None' = None, media_type: 'str | None' = None, create_time: 'str | None' = None, update_time: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneFileSearchDocument(name: 'str', display_name: 'str | None' = None, custom_metadata: 'list[dict[str, Any]]' = <factory>, state: 'str | None' = None, size_bytes: 'int | None' = None, media_type: 'str | None' = None, create_time: 'str | None' = None, update_time: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)FileSearchDocumentListResult
Beta · types
from zhivex_ai import FileSearchDocumentListResultFileSearchDocumentListResult(documents: 'list[FileSearchDocument]' = <factory>, next_page_token: 'str | None' = None, raw_response: 'Any' = None) -> NoneFileSearchDocumentListResult(documents: 'list[FileSearchDocument]' = <factory>, next_page_token: 'str | None' = None, raw_response: 'Any' = None)FileSearchOperation
Beta · types
from zhivex_ai import FileSearchOperationFileSearchOperation(name: 'str', done: 'bool' = False, metadata: 'dict[str, Any]' = <factory>, response: 'dict[str, Any] | None' = None, error: 'dict[str, Any] | None' = None, raw_response: 'Any' = None) -> NoneFileSearchOperation(name: 'str', done: 'bool' = False, metadata: 'dict[str, Any]' = <factory>, response: 'dict[str, Any] | None' = None, error: 'dict[str, Any] | None' = None, raw_response: 'Any' = None)FileSearchSearchResult
Beta · types
from zhivex_ai import FileSearchSearchResultFileSearchSearchResult(results: 'list[dict[str, Any]]' = <factory>, raw_response: 'Any' = None) -> NoneFileSearchSearchResult(results: 'list[dict[str, Any]]' = <factory>, raw_response: 'Any' = None)FileSearchStore
Beta · types
from zhivex_ai import FileSearchStoreFileSearchStore(name: 'str', display_name: 'str | None' = None, create_time: 'str | None' = None, update_time: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneFileSearchStore(name: 'str', display_name: 'str | None' = None, create_time: 'str | None' = None, update_time: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)FileSearchStoreListResult
Beta · types
from zhivex_ai import FileSearchStoreListResultFileSearchStoreListResult(stores: 'list[FileSearchStore]' = <factory>, next_page_token: 'str | None' = None, raw_response: 'Any' = None) -> NoneFileSearchStoreListResult(stores: 'list[FileSearchStore]' = <factory>, next_page_token: 'str | None' = None, raw_response: 'Any' = None)FileSearchStoresClient
Beta · types
from zhivex_ai import FileSearchStoresClientFileSearchStoresClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...FilesClient
Beta · types
from zhivex_ai import FilesClientFilesClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...FinishReason
Stable · types
from zhivex_ai import FinishReasonFinishReason(*args, **kwargs)GatewayAttempt
Stable · gateway
from zhivex_ai import GatewayAttemptGatewayAttempt(provider: 'GatewayProviderId', model_id: 'str', ok: 'bool', latency_ms: 'int', error_message: 'str | None' = None, retryable: 'bool' = False, reason: '_GatewayAttemptReason | None' = None, error_type: '_GatewayAttemptErrorType | None' = None) -> NoneGatewayAttempt(provider: 'GatewayProviderId', model_id: 'str', ok: 'bool', latency_ms: 'int', error_message: 'str | None' = None, retryable: 'bool' = False, reason: '_GatewayAttemptReason | None' = None, error_type: '_GatewayAttemptErrorType | None' = None)GatewayConfig
Stable · gateway
from zhivex_ai import GatewayConfigGatewayConfig(adapters: 'dict[GatewayProviderId, Any]', model_catalog: 'ModelCatalog | None' = None, provider_costs_per_1k_tokens: 'dict[GatewayProviderId, float]' = <factory>, latency_bias_ms: 'dict[GatewayProviderId, int]' = <factory>, max_retries: 'int' = 2, attempt_timeout_ms: 'int' = 20000, attempt_timeouts_ms: 'dict[GatewayProviderId, int]' = <factory>, retry_backoff_ms: 'int' = 200, fail_on_missing_adapter: 'bool' = False, fallback_on_refusal: 'bool' = False, on_attempt: 'Any' = None, model_costs_per_1k_tokens: 'dict[GatewayProviderId, dict[str, float]]' = <factory>) -> NoneGatewayConfig(adapters: 'dict[GatewayProviderId, Any]', model_catalog: 'ModelCatalog | None' = None, provider_costs_per_1k_tokens: 'dict[GatewayProviderId, float]' = <factory>, latency_bias_ms: 'dict[GatewayProviderId, int]' = <factory>, max_retries: 'int' = 2, attempt_timeout_ms: 'int' = 20000, attempt_timeouts_ms: 'dict[GatewayProviderId, int]' = <factory>, retry_backoff_ms: 'int' = 200, fail_on_missing_adapter: 'bool' = False, fallback_on_refusal: 'bool' = False, on_attempt: 'Any' = None, model_costs_per_1k_tokens: 'dict[GatewayProviderId, dict[str, float]]' = <factory>)GatewayError
Stable · gateway
from zhivex_ai import GatewayErrorGatewayError(message: 'str', retryable: 'bool') -> 'None'Common base class for all non-exit exceptions.GatewayImageAttachment
Stable · gateway
from zhivex_ai import GatewayImageAttachmentGatewayImageAttachment(data_url: 'str', mime_type: 'str') -> NoneGatewayImageAttachment(data_url: 'str', mime_type: 'str')GatewayMessage
Stable · gateway
from zhivex_ai import GatewayMessageGatewayMessage(role: "Literal['user', 'assistant']", content: 'str', images: 'list[GatewayImageAttachment]' = <factory>) -> NoneGatewayMessage(role: "Literal['user', 'assistant']", content: 'str', images: 'list[GatewayImageAttachment]' = <factory>)GatewayModelTarget
Stable · gateway
from zhivex_ai import GatewayModelTargetGatewayModelTarget(provider: 'GatewayProviderId', model_id: 'str') -> NoneGatewayModelTarget(provider: 'GatewayProviderId', model_id: 'str')GatewayObjectResponse
Stable · gateway
from zhivex_ai import GatewayObjectResponseGatewayObjectResponse(text: 'str', provider_used: 'GatewayProviderId', model_used: 'str', latency_ms: 'int', attempts: 'list[GatewayAttempt]', usage: 'TokenUsage', usage_estimated: 'bool', route_decision: 'GatewayRouteDecision', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, object: 'Any' = None, object_mode: "Literal['native', 'prompted']" = 'prompted') -> NoneGatewayObjectResponse(text: 'str', provider_used: 'GatewayProviderId', model_used: 'str', latency_ms: 'int', attempts: 'list[GatewayAttempt]', usage: 'TokenUsage', usage_estimated: 'bool', route_decision: 'GatewayRouteDecision', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, object: 'Any' = None, object_mode: "Literal['native', 'prompted']" = 'prompted')GatewayResponse
Stable · gateway
from zhivex_ai import GatewayResponseGatewayResponse(text: 'str', provider_used: 'GatewayProviderId', model_used: 'str', latency_ms: 'int', attempts: 'list[GatewayAttempt]', usage: 'TokenUsage', usage_estimated: 'bool', route_decision: 'GatewayRouteDecision', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None) -> NoneGatewayResponse(text: 'str', provider_used: 'GatewayProviderId', model_used: 'str', latency_ms: 'int', attempts: 'list[GatewayAttempt]', usage: 'TokenUsage', usage_estimated: 'bool', route_decision: 'GatewayRouteDecision', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None)GenerateGroundedTextOutput
Stable · types
from zhivex_ai import GenerateGroundedTextOutputGenerateGroundedTextOutput(text: 'str', sources: 'list[GroundingSource]' = <factory>, queries: 'list[str]' = <factory>, supports: 'list[GroundingSupport]' = <factory>, search_entry_point: 'dict[str, Any] | None' = None, finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, messages: 'list[ModelMessage]' = <factory>, raw_response: 'Any' = None) -> NoneGenerateGroundedTextOutput(text: 'str', sources: 'list[GroundingSource]' = <factory>, queries: 'list[str]' = <factory>, supports: 'list[GroundingSupport]' = <factory>, search_entry_point: 'dict[str, Any] | None' = None, finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, messages: 'list[ModelMessage]' = <factory>, raw_response: 'Any' = None)GenerateObjectOutput
Stable · types
from zhivex_ai import GenerateObjectOutputGenerateObjectOutput(text: 'str', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, steps: 'list[GenerateTextStep]' = <factory>, messages: 'list[ModelMessage]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, object: 'Any' = None, object_mode: "Literal['native', 'prompted']" = 'prompted') -> NoneGenerateObjectOutput(text: 'str', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, steps: 'list[GenerateTextStep]' = <factory>, messages: 'list[ModelMessage]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>, object: 'Any' = None, object_mode: "Literal['native', 'prompted']" = 'prompted')GenerateResult
Beta · types
from zhivex_ai import GenerateResultGenerateResult(message: 'ModelMessage | None' = None, messages: 'list[ModelMessage] | None' = None, text: 'str | None' = None, finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, raw_response: 'Any' = None) -> NoneGenerateResult(message: 'ModelMessage | None' = None, messages: 'list[ModelMessage] | None' = None, text: 'str | None' = None, finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, raw_response: 'Any' = None)GenerateTextOutput
Stable · types
from zhivex_ai import GenerateTextOutputGenerateTextOutput(text: 'str', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, steps: 'list[GenerateTextStep]' = <factory>, messages: 'list[ModelMessage]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>) -> NoneGenerateTextOutput(text: 'str', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None, steps: 'list[GenerateTextStep]' = <factory>, messages: 'list[ModelMessage]' = <factory>, tool_results: 'list[ToolExecutionResult]' = <factory>)GeneratedMedia
Beta · types
from zhivex_ai import GeneratedMediaGeneratedMedia(provider: 'str', data: 'bytes | None' = None, b64_data: 'str | None' = None, url: 'str | None' = None, file_uri: 'str | None' = None, media_type: 'str | None' = None, text: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneGeneratedMedia(provider: 'str', data: 'bytes | None' = None, b64_data: 'str | None' = None, url: 'str | None' = None, file_uri: 'str | None' = None, media_type: 'str | None' = None, text: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)GraphWorkflow
Stable · workflow
from zhivex_ai import GraphWorkflowGraphWorkflow(*, name: 'str', steps: 'Sequence[WorkflowStep]', edges: 'Sequence[WorkflowEdge]', definition_version: 'str' = '1', entrypoints: 'Sequence[str] | None' = None, checkpoint_store: 'WorkflowCheckpointStore | None' = None, run_store: 'AgentRunStore | None' = None, adapter: 'WorkflowAdapter | None' = None, interrupt_before: 'Mapping[str, str | None] | Sequence[str]' = (), interrupt_after: 'Mapping[str, str | None] | Sequence[str]' = (), max_concurrency: 'int | None' = None, lease_manager: 'WorkflowLeaseManager | None' = None, lease_ttl_ms: 'int' = 30000, lease_heartbeat_ms: 'int | None' = None, observer: 'AgentObserver | None' = None) -> 'None'Durable DAG workflow with append-only checkpoints and explicit resume.GroundedLanguageModel
Stable · types
from zhivex_ai import GroundedLanguageModelGroundedLanguageModel(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...GroundingSource
Beta · types
from zhivex_ai import GroundingSourceGroundingSource(url: 'str', title: 'str | None' = None, snippet: 'str | None' = None, kind: 'str | None' = None, provider_metadata: 'dict[str, Any] | None' = None) -> NoneGroundingSource(url: 'str', title: 'str | None' = None, snippet: 'str | None' = None, kind: 'str | None' = None, provider_metadata: 'dict[str, Any] | None' = None)GuardrailResult
Beta · agent
from zhivex_ai import GuardrailResultGuardrailResult(tripwire_triggered: 'bool' = False, reason: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneGuardrailResult(tripwire_triggered: 'bool' = False, reason: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)GuardrailTripwireTriggered
Beta · agent
from zhivex_ai import GuardrailTripwireTriggeredGuardrailTripwireTriggered(*, stage: "Literal['input', 'output']", guardrail_name: 'str', reason: 'str | None' = None, metadata: 'dict[str, Any] | None' = None) -> 'None'Unspecified run-time error.HTTPResponse
Stable · transport
from zhivex_ai import HTTPResponseHTTPResponse(body: 'str | bytes | AsyncIterable[bytes]', status_code: 'int' = 200, headers: 'dict[str, str]' = <factory>) -> NoneHTTPResponse(body: 'str | bytes | AsyncIterable[bytes]', status_code: 'int' = 200, headers: 'dict[str, str]' = <factory>)HTTPTransport
Stable · transport
from zhivex_ai import HTTPTransportHTTPTransport(*, client: 'httpx.AsyncClient | None' = None) -> 'None'An application-owned Fetcher with one connection pool and explicit shutdown.
Pass this object as ``fetch=transport`` to provider factories. Create one
per application lifespan/event loop and close it after in-flight work ends.
A supplied httpx client is borrowed and remains owned by the caller.HierarchicalAgentTrace
Beta · observability
from zhivex_ai import HierarchicalAgentTraceHierarchicalAgentTrace(root: 'HierarchicalAgentTraceNode', total_runs: 'int') -> NoneHierarchicalAgentTrace(root: 'HierarchicalAgentTraceNode', total_runs: 'int')HierarchicalAgentTraceNode
Beta · observability
from zhivex_ai import HierarchicalAgentTraceNodeHierarchicalAgentTraceNode(trace: 'AgentTraceArtifact', children: "list['HierarchicalAgentTraceNode']") -> NoneHierarchicalAgentTraceNode(trace: 'AgentTraceArtifact', children: "list['HierarchicalAgentTraceNode']")HostedAgentRunOptions
Beta · protocol
from zhivex_ai import HostedAgentRunOptionsHostedAgentRunOptions(session: 'AgentSession | None' = None, deps: 'Any' = None, idempotency_key: 'str | None' = None, runtime: 'AgentRuntime | None' = None) -> NoneTrusted, application-resolved options for one hosted agent run.
Authentication and tenant ownership remain application responsibilities.
In particular, ``deps`` is intentionally opaque and must never be persisted
by a protocol adapter.HostedToolClass
Beta · types
from zhivex_ai import HostedToolClassHostedToolClass(*args, **kwargs)HostedToolDefinition
Beta · types
from zhivex_ai import HostedToolDefinitionHostedToolDefinition(kind: "Literal['hosted']" = 'hosted', name: 'str' = '', provider: 'str | None' = None, type: 'str' = '', config: 'JsonValue | None' = None, tool_class: 'HostedToolClass | None' = None, requires_approval: 'bool | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneHostedToolDefinition(kind: "Literal['hosted']" = 'hosted', name: 'str' = '', provider: 'str | None' = None, type: 'str' = '', config: 'JsonValue | None' = None, tool_class: 'HostedToolClass | None' = None, requires_approval: 'bool | None' = None, metadata: 'dict[str, Any]' = <factory>)ImagePart
Beta · types
from zhivex_ai import ImagePartImagePart(type: "Literal['image']" = 'image', image: 'str' = '', media_type: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneImagePart(type: "Literal['image']" = 'image', image: 'str' = '', media_type: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)ImagesClient
Beta · types
from zhivex_ai import ImagesClientImagesClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...ImagesResult
Beta · types
from zhivex_ai import ImagesResultImagesResult(images: 'list[ProviderImage]' = <factory>, created_at: 'str | int | None' = None, raw_response: 'Any' = None) -> NoneImagesResult(images: 'list[ProviderImage]' = <factory>, created_at: 'str | int | None' = None, raw_response: 'Any' = None)InMemoryAgentRunStore
Stable · agent
from zhivex_ai import InMemoryAgentRunStoreInMemoryAgentRunStore() -> 'None'InMemoryResponsesEventStore
Beta · protocol
from zhivex_ai import InMemoryResponsesEventStoreInMemoryResponsesEventStore() -> 'None'Process-local reference store for tests and development.
Multi-replica deployments must supply an application-owned implementation
that scopes every operation using the trusted ``ProtocolInvocation``.InMemoryWorkflowCheckpointStore
Stable · workflow
from zhivex_ai import InMemoryWorkflowCheckpointStoreInMemoryWorkflowCheckpointStore() -> 'None'InMemoryWorkflowLeaseManager
Stable · workflow
from zhivex_ai import InMemoryWorkflowLeaseManagerInMemoryWorkflowLeaseManager() -> 'None'InputGuardrail
Beta · agent
from zhivex_ai import InputGuardrailInputGuardrail(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...InputGuardrailRequest
Beta · agent
from zhivex_ai import InputGuardrailRequestInputGuardrailRequest(run_id: 'str', session_id: 'str', agent_name: 'str', prompt: 'str | None' = None, messages: 'list[ModelMessage]' = <factory>, context: 'AgentContext | None' = None) -> NoneInputGuardrailRequest(run_id: 'str', session_id: 'str', agent_name: 'str', prompt: 'str | None' = None, messages: 'list[ModelMessage]' = <factory>, context: 'AgentContext | None' = None)InstalledSkill
Beta · skills
from zhivex_ai import InstalledSkillInstalledSkill(name: 'str', version: 'str', source: 'str', checksum: 'str', install_path: 'str', content_checksum: 'str | None' = None, manifest_path: 'str | None' = None, locked_at: 'str | None' = None) -> NoneInstalledSkill(name: 'str', version: 'str', source: 'str', checksum: 'str', install_path: 'str', content_checksum: 'str | None' = None, manifest_path: 'str | None' = None, locked_at: 'str | None' = None)InteractionsClient
Beta · types
from zhivex_ai import InteractionsClientInteractionsClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...JsonValue
Stable · types
from zhivex_ai import JsonValueJsonValue (public type alias or constant; no callable signature)KIMI_OFFICIAL_TOOL_URIS
Beta · provider
from zhivex_ai import KIMI_OFFICIAL_TOOL_URISKIMI_OFFICIAL_TOOL_URIS (public type alias or constant; no callable signature)KimiFormulaClient
Beta · provider
from zhivex_ai import KimiFormulaClientKimiFormulaClient(api_key: 'str', base_url: 'str', fetch: 'Fetcher', provider: 'str' = 'kimi') -> NoneKimiFormulaClient(api_key: 'str', base_url: 'str', fetch: 'Fetcher', provider: 'str' = 'kimi')LanguageModel
Stable · types
from zhivex_ai import LanguageModelLanguageModel(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...LiveAgentStreamResult
Experimental · agent
from zhivex_ai import LiveAgentStreamResultLiveAgentStreamResult(runner: 'asyncio.Task[AgentRunResult[AgentOutputT]]', broadcast: 'Broadcast[AgentLiveEvent]', live_session: 'asyncio.Future[Any]') -> 'None'Abstract base class for generic types.
On Python 3.12 and newer, generic classes implicitly inherit from
Generic when they declare a parameter list after the class's name::
class Mapping[KT, VT]:
def __getitem__(self, key: KT) -> VT:
...
# Etc.
On older versions of Python, however, generic classes have to
explicitly inherit from Generic.
After a class has been declared to be generic, it can then be used as
follows::
def lookup_name[KT, VT](mapping: Mapping[KT, VT], key: KT, default: VT) -> VT:
try:
return mapping[key]
except KeyError:
return defaultLoopAgent
Stable · workflow
from zhivex_ai import LoopAgentLoopAgent(*, name: 'str', steps: 'Sequence[WorkflowStep]', max_iterations: 'int', stop_condition: 'WorkflowStopCondition | None' = None, run_store: 'AgentRunStore | None' = None) -> 'None'MCPServerConfig
Beta · types
from zhivex_ai import MCPServerConfigMCPServerConfig(transport: "Literal['stdio', 'streamable-http']", name: 'str' = 'default', command: 'str | None' = None, args: 'list[str]' = <factory>, env: 'dict[str, str]' = <factory>, url: 'str | None' = None, headers: 'dict[str, str]' = <factory>, timeout_ms: 'int | None' = None) -> NoneMCPServerConfig(transport: "Literal['stdio', 'streamable-http']", name: 'str' = 'default', command: 'str | None' = None, args: 'list[str]' = <factory>, env: 'dict[str, str]' = <factory>, url: 'str | None' = None, headers: 'dict[str, str]' = <factory>, timeout_ms: 'int | None' = None)MCPToolConfig
Beta · types
from zhivex_ai import MCPToolConfigMCPToolConfig(server: 'MCPServerConfig', tool_name: 'str') -> NoneMCPToolConfig(server: 'MCPServerConfig', tool_name: 'str')MediaClient
Beta · types
from zhivex_ai import MediaClientMediaClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...MediaResult
Beta · types
from zhivex_ai import MediaResultMediaResult(media: 'list[GeneratedMedia]' = <factory>, text: 'str | None' = None, raw_response: 'Any' = None) -> NoneMediaResult(media: 'list[GeneratedMedia]' = <factory>, text: 'str | None' = None, raw_response: 'Any' = None)MessageRole
Beta · types
from zhivex_ai import MessageRoleMessageRole(*args, **kwargs)ModelApiSurface
Stable · catalog
from zhivex_ai import ModelApiSurfaceModelApiSurface(*args, **kwargs)ModelAvailability
Stable · catalog
from zhivex_ai import ModelAvailabilityModelAvailability(*args, **kwargs)ModelCapabilities
Stable · types
from zhivex_ai import ModelCapabilitiesModelCapabilities(streaming: 'bool', tools: 'bool', structured_output: 'bool', json_mode: 'bool', tool_choice: 'bool', parallel_tool_calls: 'bool', vision: 'bool', files: 'bool', audio_input: 'bool', audio_output: 'bool', embeddings: 'bool', reasoning: 'bool', web_search: 'bool', agent_capabilities: 'AgentCapabilities | None' = None, realtime: 'bool' = False, realtime_audio_input: 'bool' = False, realtime_audio_output: 'bool' = False, realtime_tools: 'bool' = False, realtime_browser_tokens: 'bool' = False) -> NoneModelCapabilities(streaming: 'bool', tools: 'bool', structured_output: 'bool', json_mode: 'bool', tool_choice: 'bool', parallel_tool_calls: 'bool', vision: 'bool', files: 'bool', audio_input: 'bool', audio_output: 'bool', embeddings: 'bool', reasoning: 'bool', web_search: 'bool', agent_capabilities: 'AgentCapabilities | None' = None, realtime: 'bool' = False, realtime_audio_input: 'bool' = False, realtime_audio_output: 'bool' = False, realtime_tools: 'bool' = False, realtime_browser_tokens: 'bool' = False)ModelCatalog
Stable · catalog
from zhivex_ai import ModelCatalogModelCatalog(entries: 'Iterable[ModelCatalogEntry]') -> 'None'ModelCatalogEntry
Stable · catalog
from zhivex_ai import ModelCatalogEntryModelCatalogEntry(provider: 'CatalogProviderId', model_id: 'str', aliases: 'Sequence[str]' = (), cost_per_1k_tokens: 'float | None' = None, recommended_for: 'Sequence[RecommendedUse]' = (), api_surface: 'ModelApiSurface' = 'language', availability: 'ModelAvailability' = 'stable', regions: 'Sequence[str]' = (), support_evidence: 'ModelSupportEvidence' = 'catalog-only', source_urls: 'Sequence[str]' = (), max_tool_calls_per_turn: 'int | None' = None, parallel_tool_calls: 'bool | None' = None, structured_output: 'bool | None' = None, capabilities: 'ModelCapabilities | None' = None, pricing: 'ModelPricing | None' = None, verified_at: 'str | None' = None, replacement_model_id: 'str | None' = None) -> NoneModelCatalogEntry(provider: 'CatalogProviderId', model_id: 'str', aliases: 'Sequence[str]' = (), cost_per_1k_tokens: 'float | None' = None, recommended_for: 'Sequence[RecommendedUse]' = (), api_surface: 'ModelApiSurface' = 'language', availability: 'ModelAvailability' = 'stable', regions: 'Sequence[str]' = (), support_evidence: 'ModelSupportEvidence' = 'catalog-only', source_urls: 'Sequence[str]' = (), max_tool_calls_per_turn: 'int | None' = None, parallel_tool_calls: 'bool | None' = None, structured_output: 'bool | None' = None, capabilities: 'ModelCapabilities | None' = None, pricing: 'ModelPricing | None' = None, verified_at: 'str | None' = None, replacement_model_id: 'str | None' = None)ModelGenerateInput
Beta · types
from zhivex_ai import ModelGenerateInputModelGenerateInput(timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, messages: 'list[ModelMessage]' = <factory>, tools: "dict[str, 'AnyToolDefinition'] | None" = None, tool_choice: 'ToolChoice | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'ProviderOptions | None' = None, structured_output: 'StructuredOutputConfig | None' = None) -> NoneModelGenerateInput(timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, messages: 'list[ModelMessage]' = <factory>, tools: "dict[str, 'AnyToolDefinition'] | None" = None, tool_choice: 'ToolChoice | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'ProviderOptions | None' = None, structured_output: 'StructuredOutputConfig | None' = None)ModelMessage
Stable · types
from zhivex_ai import ModelMessageModelMessage(role: 'MessageRole', parts: 'list[ContentPart]') -> NoneModelMessage(role: 'MessageRole', parts: 'list[ContentPart]')ModelPricing
Stable · catalog
from zhivex_ai import ModelPricingModelPricing(currency: 'str', source_url: 'str', input_per_1m_tokens: 'float | None' = None, output_per_1m_tokens: 'float | None' = None, cached_input_per_1m_tokens: 'float | None' = None, effective_from: 'str | None' = None, effective_until: 'str | None' = None) -> NoneSource-backed token pricing used to derive a conservative routing rate.ModelSupportEvidence
Stable · catalog
from zhivex_ai import ModelSupportEvidenceModelSupportEvidence(*args, **kwargs)ModerationsClient
Beta · types
from zhivex_ai import ModerationsClientModerationsClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...NativeSupport
Beta · types
from zhivex_ai import NativeSupportNativeSupport(text_generation: 'bool' = False, streaming: 'bool' = False, tools: 'bool' = False, structured_output: 'bool' = False, embeddings: 'bool' = False, grounding: 'bool' = False, transcription: 'bool' = False, speech: 'bool' = False, realtime: 'bool' = False, files: 'bool' = False, file_search: 'bool' = False, images: 'bool' = False, uploads: 'bool' = False, moderations: 'bool' = False, batches: 'bool' = False, videos: 'bool' = False, media: 'bool' = False, interactions: 'bool' = False, containers: 'bool' = False, skills: 'bool' = False, responses: 'bool' = False, conversations: 'bool' = False, caches: 'bool' = False, count_tokens: 'bool' = False, formulas: 'bool' = False, messages: 'bool' = False, agent_sessions: 'bool' = False, live: 'bool' = False) -> NoneNativeSupport(text_generation: 'bool' = False, streaming: 'bool' = False, tools: 'bool' = False, structured_output: 'bool' = False, embeddings: 'bool' = False, grounding: 'bool' = False, transcription: 'bool' = False, speech: 'bool' = False, realtime: 'bool' = False, files: 'bool' = False, file_search: 'bool' = False, images: 'bool' = False, uploads: 'bool' = False, moderations: 'bool' = False, batches: 'bool' = False, videos: 'bool' = False, media: 'bool' = False, interactions: 'bool' = False, containers: 'bool' = False, skills: 'bool' = False, responses: 'bool' = False, conversations: 'bool' = False, caches: 'bool' = False, count_tokens: 'bool' = False, formulas: 'bool' = False, messages: 'bool' = False, agent_sessions: 'bool' = False, live: 'bool' = False)OTelAgentObserver
Beta · observability
from zhivex_ai import OTelAgentObserverOTelAgentObserver(tracer: 'Any') -> 'None'OpenAIMcpApprovalRequest
Beta · types
from zhivex_ai import OpenAIMcpApprovalRequestOpenAIMcpApprovalRequest(type: "Literal['mcp_approval_request']" = 'mcp_approval_request', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '') -> NoneOpenAIMcpApprovalRequest(type: "Literal['mcp_approval_request']" = 'mcp_approval_request', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '')OpenAIMcpApprovalResponse
Beta · types
from zhivex_ai import OpenAIMcpApprovalResponseOpenAIMcpApprovalResponse(type: "Literal['mcp_approval_response']" = 'mcp_approval_response', approval_request_id: 'str' = '', approve: 'bool' = False, id: 'str | None' = None, reason: 'str | None' = None) -> NoneOpenAIMcpApprovalResponse(type: "Literal['mcp_approval_response']" = 'mcp_approval_response', approval_request_id: 'str' = '', approve: 'bool' = False, id: 'str | None' = None, reason: 'str | None' = None)OpenAIMcpCall
Beta · types
from zhivex_ai import OpenAIMcpCallOpenAIMcpCall(type: "Literal['mcp_call']" = 'mcp_call', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '', approval_request_id: 'str | None' = None, error: 'str | None' = None, output: 'str | None' = None, status: "Literal['in_progress', 'completed', 'incomplete', 'calling', 'failed'] | None" = None) -> NoneOpenAIMcpCall(type: "Literal['mcp_call']" = 'mcp_call', id: 'str' = '', arguments: 'str' = '', name: 'str' = '', server_label: 'str' = '', approval_request_id: 'str | None' = None, error: 'str | None' = None, output: 'str | None' = None, status: "Literal['in_progress', 'completed', 'incomplete', 'calling', 'failed'] | None" = None)OpenAIMcpListTools
Beta · types
from zhivex_ai import OpenAIMcpListToolsOpenAIMcpListTools(type: "Literal['mcp_list_tools']" = 'mcp_list_tools', id: 'str | None' = None, server_label: 'str | None' = None, tools: 'JsonValue | None' = None) -> NoneOpenAIMcpListTools(type: "Literal['mcp_list_tools']" = 'mcp_list_tools', id: 'str | None' = None, server_label: 'str | None' = None, tools: 'JsonValue | None' = None)OpenAIProviderData
Beta · types
from zhivex_ai import OpenAIProviderDataOpenAIProviderData (public type alias or constant; no callable signature)OpenAIResponseReference
Beta · types
from zhivex_ai import OpenAIResponseReferenceOpenAIResponseReference(response_id: 'str') -> NoneOpenAIResponseReference(response_id: 'str')OutputGuardrail
Beta · agent
from zhivex_ai import OutputGuardrailOutputGuardrail(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...OutputGuardrailRequest
Beta · agent
from zhivex_ai import OutputGuardrailRequestOutputGuardrailRequest(run_id: 'str', session_id: 'str', agent_name: 'str', text: 'str' = '', messages: 'list[ModelMessage]' = <factory>, result: 'GenerateTextOutput | None' = None, context: 'AgentContext | None' = None) -> NoneOutputGuardrailRequest(run_id: 'str', session_id: 'str', agent_name: 'str', text: 'str' = '', messages: 'list[ModelMessage]' = <factory>, result: 'GenerateTextOutput | None' = None, context: 'AgentContext | None' = None)ParallelAgent
Stable · workflow
from zhivex_ai import ParallelAgentParallelAgent(*, name: 'str', steps: 'Sequence[WorkflowStep]', run_store: 'AgentRunStore | None' = None) -> 'None'ParseError
Beta · errors
from zhivex_ai import ParseErrorParseError (public type alias or constant; no callable signature)Common base class for all non-exit exceptions.PendingApproval
Stable · agent
from zhivex_ai import PendingApprovalPendingApproval(id: 'str', name: 'str', arguments: 'JsonValue | None' = None, provider: 'str | None' = None, reason: 'str | None' = None, tool_call_id: 'str | None' = None, permissions: 'list[str]' = <factory>, source: 'str' = 'local', metadata: 'dict[str, JsonValue]' = <factory>, created_at_ms: 'int | None' = None, handoff_path: 'list[str]' = <factory>, tool_fingerprint: 'str | None' = None) -> NonePendingApproval(id: 'str', name: 'str', arguments: 'JsonValue | None' = None, provider: 'str | None' = None, reason: 'str | None' = None, tool_call_id: 'str | None' = None, permissions: 'list[str]' = <factory>, source: 'str' = 'local', metadata: 'dict[str, JsonValue]' = <factory>, created_at_ms: 'int | None' = None, handoff_path: 'list[str]' = <factory>, tool_fingerprint: 'str | None' = None)PortableDocument
Beta · types
from zhivex_ai import PortableDocumentPortableDocument(document_id: 'str', text: 'str', title: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>) -> NonePortableDocument(document_id: 'str', text: 'str', title: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>)PortableGroundingConfig
Beta · types
from zhivex_ai import PortableGroundingConfigPortableGroundingConfig(max_sources: 'int | None' = None) -> NonePortableGroundingConfig(max_sources: 'int | None' = None)PortableProviderNamespace
Beta · provider
from zhivex_ai import PortableProviderNamespacePortableProviderNamespace(name: 'str', native_adapter: 'ProviderAdapter', portable_support: 'PortableSupport') -> NonePortableProviderNamespace(name: 'str', native_adapter: 'ProviderAdapter', portable_support: 'PortableSupport')PortableProviderTier
Beta · types
from zhivex_ai import PortableProviderTierPortableProviderTier(*args, **kwargs)PortableRetrievalConfig
Beta · types
from zhivex_ai import PortableRetrievalConfigPortableRetrievalConfig(documents: 'list[PortableDocument]' = <factory>, max_documents: 'int' = 5, max_document_chars: 'int' = 4000) -> NonePortableRetrievalConfig(documents: 'list[PortableDocument]' = <factory>, max_documents: 'int' = 5, max_document_chars: 'int' = 4000)PortableSpeechConfig
Beta · types
from zhivex_ai import PortableSpeechConfigPortableSpeechConfig(voice: 'str | None' = None, audio_format: 'str | None' = None) -> NonePortableSpeechConfig(voice: 'str | None' = None, audio_format: 'str | None' = None)PortableSupport
Beta · types
from zhivex_ai import PortableSupportPortableSupport(text_generation: 'bool' = False, streaming: 'bool' = False, structured_output: 'bool' = False, tools: 'bool' = False, embeddings: 'bool' = False, grounding: 'bool' = False, retrieval: 'bool' = False, transcription: 'bool' = False, speech: 'bool' = False, portable_badge: 'bool' = False, tier: 'PortableProviderTier' = 'native-only') -> NonePortableSupport(text_generation: 'bool' = False, streaming: 'bool' = False, structured_output: 'bool' = False, tools: 'bool' = False, embeddings: 'bool' = False, grounding: 'bool' = False, retrieval: 'bool' = False, transcription: 'bool' = False, speech: 'bool' = False, portable_badge: 'bool' = False, tier: 'PortableProviderTier' = 'native-only')PortableTranscriptionConfig
Beta · types
from zhivex_ai import PortableTranscriptionConfigPortableTranscriptionConfig(prompt: 'str | None' = None, language: 'str | None' = None) -> NonePortableTranscriptionConfig(prompt: 'str | None' = None, language: 'str | None' = None)PostgresAgentRunStore
Stable · agent
from zhivex_ai import PostgresAgentRunStorePostgresAgentRunStore(dsn: 'str', *, table_prefix: 'str' = 'zhivex_agent') -> 'None'PostgresWorkflowCheckpointStore
Stable · workflow
from zhivex_ai import PostgresWorkflowCheckpointStorePostgresWorkflowCheckpointStore(dsn: 'str | None' = None, *, table_prefix: 'str' = 'zhivex_ai', namespace: 'str' = 'default', pool: 'Any | None' = None, pool_min_size: 'int' = 1, pool_max_size: 'int' = 5) -> 'None'PostgresWorkflowLeaseManager
Stable · workflow
from zhivex_ai import PostgresWorkflowLeaseManagerPostgresWorkflowLeaseManager(dsn: 'str | None' = None, *, table_prefix: 'str' = 'zhivex_ai', namespace: 'str' = 'default', pool: 'Any | None' = None, pool_min_size: 'int' = 1, pool_max_size: 'int' = 5) -> 'None'ProtocolErrorMapper
Beta · protocol
from zhivex_ai import ProtocolErrorMapperProtocolErrorMapper(*args, **kwargs)ProtocolEventCallback
Beta · protocol
from zhivex_ai import ProtocolEventCallbackProtocolEventCallback(*args, **kwargs)ProtocolInvocation
Beta · protocol
from zhivex_ai import ProtocolInvocationProtocolInvocation(protocol: 'str', action: 'str', agent_alias: 'str | None' = None, external_ids: 'dict[str, str]' = <factory>, request: 'Any' = None, payload: 'Mapping[str, Any] | None' = None) -> NoneTrusted routing context passed only to application-owned resolvers/stores.ProtocolLimits
Beta · protocol
from zhivex_ai import ProtocolLimitsProtocolLimits(max_request_bytes: 'int' = 1048576, max_alias_chars: 'int' = 128, max_identifier_chars: 'int' = 256, max_messages: 'int' = 128, max_parts_per_message: 'int' = 64, max_text_chars: 'int' = 262144) -> NoneFinite local parsing limits; not a rate-limit or tenancy policy.ProtocolRunOptionsResolver
Beta · protocol
from zhivex_ai import ProtocolRunOptionsResolverProtocolRunOptionsResolver(*args, **kwargs)ProviderBundle
Beta · provider
from zhivex_ai import ProviderBundleProviderBundle(name: 'str', portable: 'PortableProviderNamespace', native: 'ProviderAdapter', portable_support: 'PortableSupport', native_support: 'NativeSupport', agent_capabilities: 'AgentCapabilities', tier: 'str') -> NoneProviderBundle(name: 'str', portable: 'PortableProviderNamespace', native: 'ProviderAdapter', portable_support: 'PortableSupport', native_support: 'NativeSupport', agent_capabilities: 'AgentCapabilities', tier: 'str')ProviderDataPart
Beta · types
from zhivex_ai import ProviderDataPartProviderDataPart(type: "Literal['provider-data']" = 'provider-data', provider: 'str' = '', data: 'Any' = None) -> NoneProviderDataPart(type: "Literal['provider-data']" = 'provider-data', provider: 'str' = '', data: 'Any' = None)ProviderFile
Beta · types
from zhivex_ai import ProviderFileProviderFile(provider: 'str', id: 'str', filename: 'str | None' = None, media_type: 'str | None' = None, size_bytes: 'int | None' = None, status: 'str | None' = None, url: 'str | None' = None, file_uri: 'str | None' = None, created_at: 'str | int | None' = None, downloadable: 'bool | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneProviderFile(provider: 'str', id: 'str', filename: 'str | None' = None, media_type: 'str | None' = None, size_bytes: 'int | None' = None, status: 'str | None' = None, url: 'str | None' = None, file_uri: 'str | None' = None, created_at: 'str | int | None' = None, downloadable: 'bool | None' = None, metadata: 'dict[str, Any]' = <factory>)ProviderHTTPError
Stable · errors
from zhivex_ai import ProviderHTTPErrorProviderHTTPError(message: 'str', status: 'int', *, response_body: 'str | None' = None, response_headers: 'dict[str, Any] | None' = None, retry_after_ms: 'int | None' = None, retryable: 'bool | None' = None) -> 'None'Common base class for all non-exit exceptions.ProviderImage
Beta · types
from zhivex_ai import ProviderImageProviderImage(provider: 'str', b64_json: 'str | None' = None, url: 'str | None' = None, revised_prompt: 'str | None' = None, media_type: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneProviderImage(provider: 'str', b64_json: 'str | None' = None, url: 'str | None' = None, revised_prompt: 'str | None' = None, media_type: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)ProviderSupportRow
Beta · provider-support
from zhivex_ai import ProviderSupportRowProviderSupportRow(provider: 'str', tier: 'PortableProviderTier', portable_badge: 'bool', portable_support: 'PortableSupport', native_support: 'NativeSupport', agent_capabilities: 'AgentCapabilities', evidence_status: 'ProviderEvidenceStatus' = 'experimental/native-only') -> NoneProviderSupportRow(provider: 'str', tier: 'PortableProviderTier', portable_badge: 'bool', portable_support: 'PortableSupport', native_support: 'NativeSupport', agent_capabilities: 'AgentCapabilities', evidence_status: 'ProviderEvidenceStatus' = 'experimental/native-only')ProviderUpload
Beta · types
from zhivex_ai import ProviderUploadProviderUpload(provider: 'str', id: 'str', filename: 'str | None' = None, purpose: 'str | None' = None, bytes: 'int | None' = None, status: 'str | None' = None, mime_type: 'str | None' = None, created_at: 'str | int | None' = None, expires_at: 'str | int | None' = None, completed_at: 'str | int | None' = None, cancelled_at: 'str | int | None' = None, file: 'ProviderFile | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneProviderUpload(provider: 'str', id: 'str', filename: 'str | None' = None, purpose: 'str | None' = None, bytes: 'int | None' = None, status: 'str | None' = None, mime_type: 'str | None' = None, created_at: 'str | int | None' = None, expires_at: 'str | int | None' = None, completed_at: 'str | int | None' = None, cancelled_at: 'str | int | None' = None, file: 'ProviderFile | None' = None, metadata: 'dict[str, Any]' = <factory>)ProviderUploadPart
Beta · types
from zhivex_ai import ProviderUploadPartProviderUploadPart(provider: 'str', id: 'str', upload_id: 'str | None' = None, created_at: 'str | int | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneProviderUploadPart(provider: 'str', id: 'str', upload_id: 'str | None' = None, created_at: 'str | int | None' = None, metadata: 'dict[str, Any]' = <factory>)RealtimeAudioOutputEvent
Experimental · types
from zhivex_ai import RealtimeAudioOutputEventRealtimeAudioOutputEvent(type: "Literal['realtime-audio-output']" = 'realtime-audio-output', audio: 'bytes' = b'', media_type: 'str' = 'audio/pcm', sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeAudioOutputEvent(type: "Literal['realtime-audio-output']" = 'realtime-audio-output', audio: 'bytes' = b'', media_type: 'str' = 'audio/pcm', sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)RealtimeConnectOptions
Experimental · types
from zhivex_ai import RealtimeConnectOptionsRealtimeConnectOptions(timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, metadata: 'dict[str, Any]' = <factory>, browser_client: 'bool' = False, subprotocols: 'list[str]' = <factory>) -> NoneRealtimeConnectOptions(timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, metadata: 'dict[str, Any]' = <factory>, browser_client: 'bool' = False, subprotocols: 'list[str]' = <factory>)RealtimeErrorEvent
Experimental · types
from zhivex_ai import RealtimeErrorEventRealtimeErrorEvent(type: "Literal['realtime-error']" = 'realtime-error', error: 'Exception | None' = None, message: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeErrorEvent(type: "Literal['realtime-error']" = 'realtime-error', error: 'Exception | None' = None, message: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)RealtimeEvent
Experimental · types
from zhivex_ai import RealtimeEventRealtimeEvent (public type alias or constant; no callable signature)RealtimeModel
Experimental · types
from zhivex_ai import RealtimeModelRealtimeModel(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...RealtimeResponseCompletedEvent
Experimental · types
from zhivex_ai import RealtimeResponseCompletedEventRealtimeResponseCompletedEvent(type: "Literal['realtime-response-complete']" = 'realtime-response-complete', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeResponseCompletedEvent(type: "Literal['realtime-response-complete']" = 'realtime-response-complete', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)RealtimeSession
Experimental · types
from zhivex_ai import RealtimeSessionRealtimeSession(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...RealtimeSessionConfig
Experimental · types
from zhivex_ai import RealtimeSessionConfigRealtimeSessionConfig(instructions: 'str | None' = None, voice: 'str | None' = None, tools: "dict[str, 'AnyToolDefinition'] | None" = None, tool_choice: 'ToolChoice | None' = None, input_audio_media_type: 'str | None' = None, output_audio_media_type: 'str | None' = None, input_sample_rate_hz: 'int | None' = None, output_sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, translation_target_language_code: 'str | None' = None, translation_echo_target_language: 'bool | None' = None, turn_detection: 'dict[str, Any] | None' = None, provider_options: 'ProviderOptions | None' = None, metadata: 'dict[str, Any]' = <factory>, auto_response: 'bool' = True) -> NoneRealtimeSessionConfig(instructions: 'str | None' = None, voice: 'str | None' = None, tools: "dict[str, 'AnyToolDefinition'] | None" = None, tool_choice: 'ToolChoice | None' = None, input_audio_media_type: 'str | None' = None, output_audio_media_type: 'str | None' = None, input_sample_rate_hz: 'int | None' = None, output_sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, translation_target_language_code: 'str | None' = None, translation_echo_target_language: 'bool | None' = None, turn_detection: 'dict[str, Any] | None' = None, provider_options: 'ProviderOptions | None' = None, metadata: 'dict[str, Any]' = <factory>, auto_response: 'bool' = True)RealtimeSessionEndedEvent
Experimental · types
from zhivex_ai import RealtimeSessionEndedEventRealtimeSessionEndedEvent(type: "Literal['realtime-end']" = 'realtime-end', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeSessionEndedEvent(type: "Literal['realtime-end']" = 'realtime-end', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)RealtimeSessionStartedEvent
Experimental · types
from zhivex_ai import RealtimeSessionStartedEventRealtimeSessionStartedEvent(type: "Literal['realtime-start']" = 'realtime-start', session_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeSessionStartedEvent(type: "Literal['realtime-start']" = 'realtime-start', session_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)RealtimeTextDeltaEvent
Experimental · types
from zhivex_ai import RealtimeTextDeltaEventRealtimeTextDeltaEvent(type: "Literal['realtime-text-delta']" = 'realtime-text-delta', text_delta: 'str' = '', item_id: 'str | None' = None, response_id: 'str | None' = None, role: "Literal['assistant']" = 'assistant', provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeTextDeltaEvent(type: "Literal['realtime-text-delta']" = 'realtime-text-delta', text_delta: 'str' = '', item_id: 'str | None' = None, response_id: 'str | None' = None, role: "Literal['assistant']" = 'assistant', provider_metadata: 'dict[str, Any]' = <factory>)RealtimeTokenResult
Experimental · types
from zhivex_ai import RealtimeTokenResultRealtimeTokenResult(value: 'str', expires_at_ms: 'int | None' = None, raw_response: 'Any' = None) -> NoneRealtimeTokenResult(value: 'str', expires_at_ms: 'int | None' = None, raw_response: 'Any' = None)RealtimeToolCallEvent
Experimental · types
from zhivex_ai import RealtimeToolCallEventRealtimeToolCallEvent(type: "Literal['realtime-tool-call']" = 'realtime-tool-call', tool_call: 'ToolCall' = <factory>) -> NoneRealtimeToolCallEvent(type: "Literal['realtime-tool-call']" = 'realtime-tool-call', tool_call: 'ToolCall' = <factory>)RealtimeToolResultEvent
Experimental · types
from zhivex_ai import RealtimeToolResultEventRealtimeToolResultEvent(type: "Literal['realtime-tool-result']" = 'realtime-tool-result', tool_result: 'ToolExecutionResult' = <factory>) -> NoneRealtimeToolResultEvent(type: "Literal['realtime-tool-result']" = 'realtime-tool-result', tool_result: 'ToolExecutionResult' = <factory>)RealtimeTranscriptEvent
Experimental · types
from zhivex_ai import RealtimeTranscriptEventRealtimeTranscriptEvent(type: "Literal['realtime-transcript']" = 'realtime-transcript', text: 'str' = '', role: "Literal['user', 'assistant']" = 'assistant', is_final: 'bool' = False, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneRealtimeTranscriptEvent(type: "Literal['realtime-transcript']" = 'realtime-transcript', text: 'str' = '', role: "Literal['user', 'assistant']" = 'assistant', is_final: 'bool' = False, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)ReasoningConfig
Beta · types
from zhivex_ai import ReasoningConfigReasoningConfig(effort: "Literal['none', 'minimal', 'low', 'medium', 'high', 'xhigh', 'max'] | None" = None, budget_tokens: 'int | None' = None) -> NoneReasoningConfig(effort: "Literal['none', 'minimal', 'low', 'medium', 'high', 'xhigh', 'max'] | None" = None, budget_tokens: 'int | None' = None)RecommendedUse
Stable · catalog
from zhivex_ai import RecommendedUseRecommendedUse(*args, **kwargs)RedactionPolicy
Beta · safety
from zhivex_ai import RedactionPolicyRedactionPolicy(rules: 'list[RedactionRule]') -> NoneRedactionPolicy(rules: 'list[RedactionRule]')RedactionRule
Beta · safety
from zhivex_ai import RedactionRuleRedactionRule(pattern: 'str | re.Pattern[str]', replacement: 'str' = '[REDACTED]', name: 'str | None' = None) -> NoneRedactionRule(pattern: 'str | re.Pattern[str]', replacement: 'str' = '[REDACTED]', name: 'str | None' = None)RemoteHTTPToolConfig
Beta · types
from zhivex_ai import RemoteHTTPToolConfigRemoteHTTPToolConfig(url: 'str', headers: 'dict[str, str]' = <factory>, timeout_ms: 'int | None' = None) -> NoneRemoteHTTPToolConfig(url: 'str', headers: 'dict[str, str]' = <factory>, timeout_ms: 'int | None' = None)ResponsesAgentHost
Beta · protocol
from zhivex_ai import ResponsesAgentHostResponsesAgentHost(agents: 'AgentResolver', *, run_options_resolver: 'ProtocolRunOptionsResolver | None' = None, error_mapper: 'ProtocolErrorMapper | None' = None, on_protocol_event: 'ProtocolEventCallback | None' = None, limits: 'ProtocolLimits | None' = None, event_store: 'ResponsesEventStore | None' = None) -> 'None'Host configured agents behind the OpenAI Responses create/stream shape.
The request ``model`` is an application-owned alias. It never accepts or
constructs arbitrary provider credentials or provider model identifiers.ResponsesClient
Beta · types
from zhivex_ai import ResponsesClientResponsesClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...ResponsesEventStore
Beta · protocol
from zhivex_ai import ResponsesEventStoreResponsesEventStore(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...RunLimits
Beta · agent
from zhivex_ai import RunLimitsRunLimits(max_steps: 'int | None' = 8, max_tool_calls: 'int | None' = 32, max_wall_time_ms: 'int | None' = 120000, max_handoffs: 'int | None' = 1) -> NoneRunLimits(max_steps: 'int | None' = 8, max_tool_calls: 'int | None' = 32, max_wall_time_ms: 'int | None' = 120000, max_handoffs: 'int | None' = 1)SQLiteAgentRunStore
Stable · agent
from zhivex_ai import SQLiteAgentRunStoreSQLiteAgentRunStore(path: 'str', *, namespace: 'str' = 'default') -> 'None'SQLiteWorkflowCheckpointStore
Stable · workflow
from zhivex_ai import SQLiteWorkflowCheckpointStoreSQLiteWorkflowCheckpointStore(path: 'str', *, namespace: 'str' = 'default') -> 'None'SQLiteWorkflowLeaseManager
Stable · workflow
from zhivex_ai import SQLiteWorkflowLeaseManagerSQLiteWorkflowLeaseManager(path: 'str', *, namespace: 'str' = 'default') -> 'None'SafetyPolicy
Beta · safety
from zhivex_ai import SafetyPolicySafetyPolicy(preset: 'SafetyPolicyPreset', approval_policy: 'ApprovalPolicy | None' = None, input_guardrails: 'list[InputGuardrail]' = <factory>, output_guardrails: 'list[OutputGuardrail]' = <factory>, tool_execution: 'ToolExecutionOptions | None' = None, redaction: 'RedactionPolicy | None' = None, budget: 'BudgetGuard | None' = None, run_limits: 'RunLimits | None' = None) -> NoneSafetyPolicy(preset: 'SafetyPolicyPreset', approval_policy: 'ApprovalPolicy | None' = None, input_guardrails: 'list[InputGuardrail]' = <factory>, output_guardrails: 'list[OutputGuardrail]' = <factory>, tool_execution: 'ToolExecutionOptions | None' = None, redaction: 'RedactionPolicy | None' = None, budget: 'BudgetGuard | None' = None, run_limits: 'RunLimits | None' = None)SafetyPolicyPreset
Beta · safety
from zhivex_ai import SafetyPolicyPresetSafetyPolicyPreset(*args, **kwargs)SequentialAgent
Stable · workflow
from zhivex_ai import SequentialAgentSequentialAgent(*, name: 'str', steps: 'Sequence[WorkflowStep]', run_store: 'AgentRunStore | None' = None) -> 'None'SkillArtifact
Beta · skills
from zhivex_ai import SkillArtifactSkillArtifact(name: 'str', path: 'str', media_type: 'str | None' = None, role: 'SkillArtifactRole' = 'primary', description: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneSkillArtifact(name: 'str', path: 'str', media_type: 'str | None' = None, role: 'SkillArtifactRole' = 'primary', description: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)SkillDefinition
Stable · skills
from zhivex_ai import SkillDefinitionSkillDefinition(name: 'str', description: 'str', instructions: 'str', path: 'str | None' = None, metadata_path: 'str | None' = None, display_name: 'str | None' = None, short_description: 'str | None' = None, default_prompt: 'str | None' = None, allow_implicit_invocation: 'bool' = True, priority: 'int' = 0, triggers: 'list[str]' = <factory>, anti_triggers: 'list[str]' = <factory>, allowed_providers: 'list[str]' = <factory>, allowed_models: 'list[str]' = <factory>, persist_to_session: 'bool' = True, dependency_failure_mode: 'SkillDependencyFailureMode' = 'skip', tools: 'dict[str, ToolDefinition]' = <factory>, dependencies: 'list[SkillDependency]' = <factory>, metadata: 'dict[str, Any]' = <factory>, version: 'str | None' = None, entrypoints: 'list[SkillEntrypoint]' = <factory>, artifacts: 'list[SkillArtifact]' = <factory>, permissions: 'SkillPermissions' = <factory>, resources: 'list[str]' = <factory>, source: 'str | None' = None, checksum: 'str | None' = None, package_manifest: 'SkillPackageManifest | None' = None, package_manifest_path: 'str | None' = None, install_path: 'str | None' = None) -> NoneSkillDefinition(name: 'str', description: 'str', instructions: 'str', path: 'str | None' = None, metadata_path: 'str | None' = None, display_name: 'str | None' = None, short_description: 'str | None' = None, default_prompt: 'str | None' = None, allow_implicit_invocation: 'bool' = True, priority: 'int' = 0, triggers: 'list[str]' = <factory>, anti_triggers: 'list[str]' = <factory>, allowed_providers: 'list[str]' = <factory>, allowed_models: 'list[str]' = <factory>, persist_to_session: 'bool' = True, dependency_failure_mode: 'SkillDependencyFailureMode' = 'skip', tools: 'dict[str, ToolDefinition]' = <factory>, dependencies: 'list[SkillDependency]' = <factory>, metadata: 'dict[str, Any]' = <factory>, version: 'str | None' = None, entrypoints: 'list[SkillEntrypoint]' = <factory>, artifacts: 'list[SkillArtifact]' = <factory>, permissions: 'SkillPermissions' = <factory>, resources: 'list[str]' = <factory>, source: 'str | None' = None, checksum: 'str | None' = None, package_manifest: 'SkillPackageManifest | None' = None, package_manifest_path: 'str | None' = None, install_path: 'str | None' = None)SkillDependency
Stable · skills
from zhivex_ai import SkillDependencySkillDependency(type: 'SkillDependencyType', value: 'str', description: 'str | None' = None, transport: 'SkillTransport' = 'streamable-http', url: 'str | None' = None, headers: 'dict[str, str]' = <factory>, timeout_ms: 'int | None' = None, command: 'str | None' = None, args: 'list[str]' = <factory>, env: 'dict[str, str]' = <factory>, include: 'list[str]' = <factory>, exclude: 'list[str]' = <factory>, prefix: 'str | None' = None, version: 'str | None' = None, required: 'bool' = True, import_name: 'str | None' = None) -> NoneSkillDependency(type: 'SkillDependencyType', value: 'str', description: 'str | None' = None, transport: 'SkillTransport' = 'streamable-http', url: 'str | None' = None, headers: 'dict[str, str]' = <factory>, timeout_ms: 'int | None' = None, command: 'str | None' = None, args: 'list[str]' = <factory>, env: 'dict[str, str]' = <factory>, include: 'list[str]' = <factory>, exclude: 'list[str]' = <factory>, prefix: 'str | None' = None, version: 'str | None' = None, required: 'bool' = True, import_name: 'str | None' = None)SkillEntrypoint
Beta · skills
from zhivex_ai import SkillEntrypointSkillEntrypoint(name: 'str', description: 'str | None' = None, runtime: 'SkillEntrypointRuntime' = 'python', script: 'str | None' = None, default: 'bool' = False, input_schema: 'dict[str, Any]' = <factory>, tool_name: 'str | None' = None) -> NoneSkillEntrypoint(name: 'str', description: 'str | None' = None, runtime: 'SkillEntrypointRuntime' = 'python', script: 'str | None' = None, default: 'bool' = False, input_schema: 'dict[str, Any]' = <factory>, tool_name: 'str | None' = None)SkillPackageManifest
Beta · skills
from zhivex_ai import SkillPackageManifestSkillPackageManifest(schema_version: 'int' = 1, name: 'str' = '', version: 'str' = '', description: 'str' = '', entrypoints: 'list[SkillEntrypoint]' = <factory>, dependencies: 'list[SkillDependency]' = <factory>, artifacts: 'list[SkillArtifact]' = <factory>, permissions: 'SkillPermissions' = <factory>, resources: 'list[str]' = <factory>, provider_overrides: 'dict[str, Any]' = <factory>) -> NoneSkillPackageManifest(schema_version: 'int' = 1, name: 'str' = '', version: 'str' = '', description: 'str' = '', entrypoints: 'list[SkillEntrypoint]' = <factory>, dependencies: 'list[SkillDependency]' = <factory>, artifacts: 'list[SkillArtifact]' = <factory>, permissions: 'SkillPermissions' = <factory>, resources: 'list[str]' = <factory>, provider_overrides: 'dict[str, Any]' = <factory>)SkillPermissions
Beta · skills
from zhivex_ai import SkillPermissionsSkillPermissions(allow_network: 'bool' = False, read_paths: 'list[str]' = <factory>, write_paths: 'list[str]' = <factory>) -> NoneSkillPermissions(allow_network: 'bool' = False, read_paths: 'list[str]' = <factory>, write_paths: 'list[str]' = <factory>)SkillRegistry
Stable · skills
from zhivex_ai import SkillRegistrySkillRegistry(skills: 'SkillSet | None' = None) -> 'None'SkillRegistryIndex
Beta · skills
from zhivex_ai import SkillRegistryIndexSkillRegistryIndex(registry_url: 'str | None' = None, skills: 'dict[str, dict[str, dict[str, Any]]]' = <factory>) -> NoneSkillRegistryIndex(registry_url: 'str | None' = None, skills: 'dict[str, dict[str, dict[str, Any]]]' = <factory>)SkillRunResult
Beta · skills
from zhivex_ai import SkillRunResultSkillRunResult(skill_name: 'str', skill_version: 'str | None' = None, entrypoint: 'str | None' = None, output: 'Any' = None, artifacts: 'list[SkillArtifact]' = <factory>, logs: 'list[str]' = <factory>) -> NoneSkillRunResult(skill_name: 'str', skill_version: 'str | None' = None, entrypoint: 'str | None' = None, output: 'Any' = None, artifacts: 'list[SkillArtifact]' = <factory>, logs: 'list[str]' = <factory>)SkillsClient
Beta · types
from zhivex_ai import SkillsClientSkillsClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...SpeechModel
Beta · types
from zhivex_ai import SpeechModelSpeechModel(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...SpeechOutput
Beta · types
from zhivex_ai import SpeechOutputSpeechOutput(audio: 'bytes', media_type: 'str', input: 'str | None' = None, raw_response: 'Any' = None) -> NoneSpeechOutput(audio: 'bytes', media_type: 'str', input: 'str | None' = None, raw_response: 'Any' = None)StoredResponsesRun
Beta · protocol
from zhivex_ai import StoredResponsesRunStoredResponsesRun(response_id: 'str', model: 'str', status: 'str' = 'in_progress', response: 'dict[str, Any] | None' = None, events: 'list[dict[str, Any]]' = <factory>, internal_run_id: 'str | None' = None) -> NoneStoredResponsesRun(response_id: 'str', model: 'str', status: 'str' = 'in_progress', response: 'dict[str, Any] | None' = None, events: 'list[dict[str, Any]]' = <factory>, internal_run_id: 'str | None' = None)StreamEvent
Stable · types
from zhivex_ai import StreamEventStreamEvent (public type alias or constant; no callable signature)StreamFinishEvent
Beta · types
from zhivex_ai import StreamFinishEventStreamFinishEvent(type: "Literal['finish']" = 'finish', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None) -> NoneStreamFinishEvent(type: "Literal['finish']" = 'finish', finish_reason: 'FinishReason | None' = None, provider_finish_reason: 'str | None' = None, usage: 'TokenUsage | None' = None)StreamObjectResult
Stable · types
from zhivex_ai import StreamObjectResultStreamObjectResult(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...StreamProviderDataEvent
Beta · types
from zhivex_ai import StreamProviderDataEventStreamProviderDataEvent(type: "Literal['provider-data']" = 'provider-data', provider: 'str' = '', data: 'Any' = None) -> NoneStreamProviderDataEvent(type: "Literal['provider-data']" = 'provider-data', provider: 'str' = '', data: 'Any' = None)StreamTextDeltaEvent
Beta · types
from zhivex_ai import StreamTextDeltaEventStreamTextDeltaEvent(type: "Literal['text-delta']" = 'text-delta', text_delta: 'str' = '') -> NoneStreamTextDeltaEvent(type: "Literal['text-delta']" = 'text-delta', text_delta: 'str' = '')StreamTextResult
Stable · types
from zhivex_ai import StreamTextResultStreamTextResult(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...SummaryConfig
Stable · agent
from zhivex_ai import SummaryConfigSummaryConfig(max_messages: 'int' = 12, preserve_recent_messages: 'int' = 8, max_summary_chars: 'int' = 1200) -> NoneSummaryConfig(max_messages: 'int' = 12, preserve_recent_messages: 'int' = 8, max_summary_chars: 'int' = 1200)TextPart
Beta · types
from zhivex_ai import TextPartTextPart(type: "Literal['text']" = 'text', text: 'str' = '', provider_metadata: 'dict[str, Any]' = <factory>) -> NoneTextPart(type: "Literal['text']" = 'text', text: 'str' = '', provider_metadata: 'dict[str, Any]' = <factory>)TokenCountDetail
Beta · types
from zhivex_ai import TokenCountDetailTokenCountDetail(modality: 'str | None' = None, token_count: 'int | None' = None, billable_characters: 'int | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneTokenCountDetail(modality: 'str | None' = None, token_count: 'int | None' = None, billable_characters: 'int | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)TokenPricing
Beta · observability
from zhivex_ai import TokenPricingTokenPricing(input_cost_per_1k_tokens: 'float | None' = None, output_cost_per_1k_tokens: 'float | None' = None, total_cost_per_1k_tokens: 'float | None' = None, currency: 'str' = 'USD') -> NoneTokenPricing(input_cost_per_1k_tokens: 'float | None' = None, output_cost_per_1k_tokens: 'float | None' = None, total_cost_per_1k_tokens: 'float | None' = None, currency: 'str' = 'USD')TokenUsage
Stable · types
from zhivex_ai import TokenUsageTokenUsage(input_tokens: 'int | None' = None, output_tokens: 'int | None' = None, total_tokens: 'int | None' = None) -> NoneTokenUsage(input_tokens: 'int | None' = None, output_tokens: 'int | None' = None, total_tokens: 'int | None' = None)ToolApprovalRequest
Stable · agent
from zhivex_ai import ToolApprovalRequestToolApprovalRequest(run_id: 'str', session_id: 'str', agent_name: 'str', tool_name: 'str', tool_input: 'Any', tool_permissions: 'list[str]' = <factory>, tool_source: 'str' = 'local', tool_metadata: 'dict[str, Any]' = <factory>, context: 'AgentContext | None' = None, handoff_path: 'list[str]' = <factory>) -> NoneToolApprovalRequest(run_id: 'str', session_id: 'str', agent_name: 'str', tool_name: 'str', tool_input: 'Any', tool_permissions: 'list[str]' = <factory>, tool_source: 'str' = 'local', tool_metadata: 'dict[str, Any]' = <factory>, context: 'AgentContext | None' = None, handoff_path: 'list[str]' = <factory>)ToolCall
Stable · types
from zhivex_ai import ToolCallToolCall(id: 'str', name: 'str', input: 'JsonValue', provider_metadata: 'dict[str, Any]' = <factory>) -> NoneToolCall(id: 'str', name: 'str', input: 'JsonValue', provider_metadata: 'dict[str, Any]' = <factory>)ToolChoiceName
Beta · types
from zhivex_ai import ToolChoiceNameToolChoiceName(tool_name: 'str') -> NoneToolChoiceName(tool_name: 'str')ToolDefinition
Stable · types
from zhivex_ai import ToolDefinitionToolDefinition(name: 'str', description: 'str | None', schema: 'Any', execute: 'Callable[..., Awaitable[JsonValue] | JsonValue] | None' = None, input_examples: 'list[JsonValue]' = <factory>, strict: 'bool | None' = None, defer_loading: 'bool | None' = None, eager_input_streaming: 'bool | None' = None, allowed_callers: 'list[str]' = <factory>, cache_control: 'dict[str, Any] | None' = None, tags: 'list[str]' = <factory>, requires_approval: 'bool | None' = None, permissions: 'list[str]' = <factory>, source: 'ToolSource' = 'local', metadata: 'dict[str, Any]' = <factory>, supports_streaming: 'bool' = False, remote_config: 'RemoteHTTPToolConfig | None' = None, mcp_config: 'MCPToolConfig | None' = None, output_schema: 'Any' = None, input_guardrails: 'list[ToolInputGuardrail]' = <factory>, output_guardrails: 'list[ToolOutputGuardrail]' = <factory>) -> NoneToolDefinition(name: 'str', description: 'str | None', schema: 'Any', execute: 'Callable[..., Awaitable[JsonValue] | JsonValue] | None' = None, input_examples: 'list[JsonValue]' = <factory>, strict: 'bool | None' = None, defer_loading: 'bool | None' = None, eager_input_streaming: 'bool | None' = None, allowed_callers: 'list[str]' = <factory>, cache_control: 'dict[str, Any] | None' = None, tags: 'list[str]' = <factory>, requires_approval: 'bool | None' = None, permissions: 'list[str]' = <factory>, source: 'ToolSource' = 'local', metadata: 'dict[str, Any]' = <factory>, supports_streaming: 'bool' = False, remote_config: 'RemoteHTTPToolConfig | None' = None, mcp_config: 'MCPToolConfig | None' = None, output_schema: 'Any' = None, input_guardrails: 'list[ToolInputGuardrail]' = <factory>, output_guardrails: 'list[ToolOutputGuardrail]' = <factory>)ToolExecutionContext
Stable · types
from zhivex_ai import ToolExecutionContextToolExecutionContext(tool_name: 'str', tool_call_id: 'str' = '', idempotency_key: 'str | None' = None, deadline_ms: 'int | None' = None, run_id: 'str | None' = None, session_id: 'str | None' = None, agent_name: 'str | None' = None, memory_summary: 'str | None' = None, permissions: 'list[str]' = <factory>, source: 'ToolSource' = 'local', metadata: 'dict[str, Any]' = <factory>, handoff_path: 'list[str]' = <factory>, deps: 'ToolContextDepsT | None' = None, cancellation_token: 'Any' = None) -> NoneToolExecutionContext(tool_name: 'str', tool_call_id: 'str' = '', idempotency_key: 'str | None' = None, deadline_ms: 'int | None' = None, run_id: 'str | None' = None, session_id: 'str | None' = None, agent_name: 'str | None' = None, memory_summary: 'str | None' = None, permissions: 'list[str]' = <factory>, source: 'ToolSource' = 'local', metadata: 'dict[str, Any]' = <factory>, handoff_path: 'list[str]' = <factory>, deps: 'ToolContextDepsT | None' = None, cancellation_token: 'Any' = None)ToolExecutionError
Stable · types
from zhivex_ai import ToolExecutionErrorToolExecutionError(message: 'str') -> NoneToolExecutionError(message: 'str')ToolExecutionOptions
Stable · types
from zhivex_ai import ToolExecutionOptionsToolExecutionOptions(parallel: 'bool | None' = None, max_concurrency: 'int | None' = None, timeout_ms: 'int | None' = None, stop_on_error: 'bool' = False) -> NoneToolExecutionOptions(parallel: 'bool | None' = None, max_concurrency: 'int | None' = None, timeout_ms: 'int | None' = None, stop_on_error: 'bool' = False)ToolExecutionOutcomeUnknown
Stable · errors
from zhivex_ai import ToolExecutionOutcomeUnknownToolExecutionOutcomeUnknown(message: 'str', *, tool_name: 'str', tool_call_id: 'str', timeout_ms: 'int', idempotency_key: 'str') -> 'None'Raised when a timed-out tool may still have produced an external side effect.ToolExecutionResult
Stable · types
from zhivex_ai import ToolExecutionResultToolExecutionResult(tool_call_id: 'str', tool_name: 'str', output: 'JsonValue | None' = None, error: 'ToolExecutionError | None' = None, is_error: 'bool' = False, provider_metadata: 'dict[str, Any]' = <factory>) -> NoneToolExecutionResult(tool_call_id: 'str', tool_name: 'str', output: 'JsonValue | None' = None, error: 'ToolExecutionError | None' = None, is_error: 'bool' = False, provider_metadata: 'dict[str, Any]' = <factory>)ToolGuardrailResult
Beta · types
from zhivex_ai import ToolGuardrailResultToolGuardrailResult(tripwire_triggered: 'bool' = False, reason: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>, replacement: 'Any' = None, replace: 'bool' = False) -> NoneToolGuardrailResult(tripwire_triggered: 'bool' = False, reason: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>, replacement: 'Any' = None, replace: 'bool' = False)ToolGuardrailStage
Beta · types
from zhivex_ai import ToolGuardrailStageToolGuardrailStage(*args, **kwargs)ToolGuardrailTripwireTriggered
Beta · types
from zhivex_ai import ToolGuardrailTripwireTriggeredToolGuardrailTripwireTriggered(*, stage: 'ToolGuardrailStage', tool_name: 'str', guardrail_name: 'str', reason: 'str | None' = None, metadata: 'dict[str, Any] | None' = None) -> 'None'Unspecified run-time error.ToolInputGuardrail
Beta · types
from zhivex_ai import ToolInputGuardrailToolInputGuardrail(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...ToolInputGuardrailRequest
Beta · types
from zhivex_ai import ToolInputGuardrailRequestToolInputGuardrailRequest(tool_name: 'str', input: 'Any', context: 'ToolExecutionContext[Any]') -> NoneToolInputGuardrailRequest(tool_name: 'str', input: 'Any', context: 'ToolExecutionContext[Any]')ToolOutputGuardrail
Beta · types
from zhivex_ai import ToolOutputGuardrailToolOutputGuardrail(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...ToolOutputGuardrailRequest
Beta · types
from zhivex_ai import ToolOutputGuardrailRequestToolOutputGuardrailRequest(tool_name: 'str', input: 'Any', output: 'Any', context: 'ToolExecutionContext[Any]') -> NoneToolOutputGuardrailRequest(tool_name: 'str', input: 'Any', output: 'Any', context: 'ToolExecutionContext[Any]')ToolRegistry
Stable · agent
from zhivex_ai import ToolRegistryToolRegistry(tools: 'ToolSet | None' = None, *, runtimes: 'dict[str, ToolRuntime] | None' = None) -> 'None'ToolSet
Stable · types
from zhivex_ai import ToolSetToolSet(*args, **kwargs)ToolSource
Beta · types
from zhivex_ai import ToolSourceToolSource(*args, **kwargs)TranscriptionModel
Beta · types
from zhivex_ai import TranscriptionModelTranscriptionModel(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...TranscriptionOutput
Beta · types
from zhivex_ai import TranscriptionOutputTranscriptionOutput(text: 'str', audio: 'AudioInput | None' = None, raw_response: 'Any' = None) -> NoneTranscriptionOutput(text: 'str', audio: 'AudioInput | None' = None, raw_response: 'Any' = None)UIMessage
Beta · types
from zhivex_ai import UIMessageUIMessage(id: 'str', role: 'MessageRole', parts: 'list[ContentPart]') -> NoneUIMessage(id: 'str', role: 'MessageRole', parts: 'list[ContentPart]')UIMessageChunk
Beta · types
from zhivex_ai import UIMessageChunkUIMessageChunk (public type alias or constant; no callable signature)UIMessageProviderDataChunk
Beta · types
from zhivex_ai import UIMessageProviderDataChunkUIMessageProviderDataChunk(type: "Literal['provider-data']" = 'provider-data', message_id: 'str' = '', role: "Literal['assistant']" = 'assistant', provider: 'str' = '', data: 'Any' = None) -> NoneUIMessageProviderDataChunk(type: "Literal['provider-data']" = 'provider-data', message_id: 'str' = '', role: "Literal['assistant']" = 'assistant', provider: 'str' = '', data: 'Any' = None)UIMessageToolApprovalChunk
Beta · types
from zhivex_ai import UIMessageToolApprovalChunkUIMessageToolApprovalChunk(type: "Literal['tool-approval']" = 'tool-approval', message_id: 'str' = '', role: "Literal['assistant']" = 'assistant', tool_name: 'str' = '', tool_input: 'JsonValue | None' = None, approved: 'bool' = False, reason: 'str | None' = None, approval_request_id: 'str | None' = None, provider: 'str | None' = None, provider_managed: 'bool' = False, tool_source: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>) -> NoneUIMessageToolApprovalChunk(type: "Literal['tool-approval']" = 'tool-approval', message_id: 'str' = '', role: "Literal['assistant']" = 'assistant', tool_name: 'str' = '', tool_input: 'JsonValue | None' = None, approved: 'bool' = False, reason: 'str | None' = None, approval_request_id: 'str | None' = None, provider: 'str | None' = None, provider_managed: 'bool' = False, tool_source: 'str | None' = None, metadata: 'dict[str, Any]' = <factory>)UnsupportedFeatureError
Stable · errors
from zhivex_ai import UnsupportedFeatureErrorUnsupportedFeatureError (public type alias or constant; no callable signature)Common base class for all non-exit exceptions.UploadsClient
Beta · types
from zhivex_ai import UploadsClientUploadsClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...ValidationError
Stable · errors
from zhivex_ai import ValidationErrorValidationError (public type alias or constant; no callable signature)Common base class for all non-exit exceptions.VideoOperation
Beta · types
from zhivex_ai import VideoOperationVideoOperation(name: 'str', done: 'bool' = False, response: 'dict[str, Any] | None' = None, error: 'dict[str, Any] | None' = None, raw_response: 'Any' = None) -> NoneVideoOperation(name: 'str', done: 'bool' = False, response: 'dict[str, Any] | None' = None, error: 'dict[str, Any] | None' = None, raw_response: 'Any' = None)VideoResult
Beta · types
from zhivex_ai import VideoResultVideoResult(videos: 'list[GeneratedMedia]' = <factory>, operation: 'VideoOperation | None' = None, raw_response: 'Any' = None) -> NoneVideoResult(videos: 'list[GeneratedMedia]' = <factory>, operation: 'VideoOperation | None' = None, raw_response: 'Any' = None)VideosClient
Beta · types
from zhivex_ai import VideosClientVideosClient(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...WORKFLOW_ADAPTER_SCHEMA_VERSION
Stable · workflow
from zhivex_ai import WORKFLOW_ADAPTER_SCHEMA_VERSIONWORKFLOW_ADAPTER_SCHEMA_VERSION (public type alias or constant; no callable signature)WORKFLOW_CHECKPOINT_SCHEMA_VERSION
Stable · workflow
from zhivex_ai import WORKFLOW_CHECKPOINT_SCHEMA_VERSIONWORKFLOW_CHECKPOINT_SCHEMA_VERSION (public type alias or constant; no callable signature)WorkflowAdapter
Stable · workflow
from zhivex_ai import WorkflowAdapterWorkflowAdapter(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...WorkflowAdapterCapabilities
Stable · workflow
from zhivex_ai import WorkflowAdapterCapabilitiesWorkflowAdapterCapabilities(durable_steps: 'bool' = False, native_step_retries: 'bool' = False, signals: 'bool' = False, explicit_resume: 'bool' = False, fork: 'bool' = False, cancellation: 'bool' = False, durable_timers: 'bool' = False) -> NoneWorkflowAdapterCapabilities(durable_steps: 'bool' = False, native_step_retries: 'bool' = False, signals: 'bool' = False, explicit_resume: 'bool' = False, fork: 'bool' = False, cancellation: 'bool' = False, durable_timers: 'bool' = False)WorkflowAgent
Stable · workflow
from zhivex_ai import WorkflowAgentWorkflowAgent(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...WorkflowBuilder
Stable · workflow
from zhivex_ai import WorkflowBuilderWorkflowBuilder(name: 'str', *, definition_version: 'str' = '1') -> 'None'Build a validated acyclic workflow graph without mutating prior builders.WorkflowCheckpoint
Stable · workflow
from zhivex_ai import WorkflowCheckpointWorkflowCheckpoint(checkpoint_id: 'str', run_id: 'str', workflow_name: 'str', definition_version: 'str', definition_digest: 'str', sequence: 'int' = 0, schema_version: 'int' = 2, status: 'WorkflowCheckpointStatus' = 'running', session_id: 'str | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, state: 'dict[str, JsonValue]' = <factory>, nodes: 'dict[str, WorkflowNodeCheckpoint]' = <factory>, edge_decisions: 'dict[str, bool]' = <factory>, ready_nodes: 'list[str]' = <factory>, pending_interrupt: 'WorkflowInterrupt | None' = None, transition: 'WorkflowTransition | None' = None, forked_from_run_id: 'str | None' = None, forked_from_checkpoint_id: 'str | None' = None, resume_values: 'dict[str, JsonValue]' = <factory>, created_at_ms: 'int | None' = None, updated_at_ms: 'int | None' = None, metadata: 'dict[str, JsonValue]' = <factory>, migration_history: 'list[WorkflowCheckpointMigration]' = <factory>) -> NoneWorkflowCheckpoint(checkpoint_id: 'str', run_id: 'str', workflow_name: 'str', definition_version: 'str', definition_digest: 'str', sequence: 'int' = 0, schema_version: 'int' = 2, status: 'WorkflowCheckpointStatus' = 'running', session_id: 'str | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, state: 'dict[str, JsonValue]' = <factory>, nodes: 'dict[str, WorkflowNodeCheckpoint]' = <factory>, edge_decisions: 'dict[str, bool]' = <factory>, ready_nodes: 'list[str]' = <factory>, pending_interrupt: 'WorkflowInterrupt | None' = None, transition: 'WorkflowTransition | None' = None, forked_from_run_id: 'str | None' = None, forked_from_checkpoint_id: 'str | None' = None, resume_values: 'dict[str, JsonValue]' = <factory>, created_at_ms: 'int | None' = None, updated_at_ms: 'int | None' = None, metadata: 'dict[str, JsonValue]' = <factory>, migration_history: 'list[WorkflowCheckpointMigration]' = <factory>)WorkflowCheckpointMigration
Stable · workflow
from zhivex_ai import WorkflowCheckpointMigrationWorkflowCheckpointMigration(migration_id: 'str', from_version: 'int', to_version: 'int', applied_at_ms: 'int', metadata: 'dict[str, JsonValue]' = <factory>) -> NoneWorkflowCheckpointMigration(migration_id: 'str', from_version: 'int', to_version: 'int', applied_at_ms: 'int', metadata: 'dict[str, JsonValue]' = <factory>)WorkflowCheckpointStatus
Stable · workflow
from zhivex_ai import WorkflowCheckpointStatusWorkflowCheckpointStatus(*args, **kwargs)WorkflowCheckpointStore
Stable · workflow
from zhivex_ai import WorkflowCheckpointStoreWorkflowCheckpointStore(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...WorkflowConflictError
Stable · errors
from zhivex_ai import WorkflowConflictErrorWorkflowConflictError (public type alias or constant; no callable signature)Raised when concurrent workflow state cannot be committed safely.WorkflowContext
Stable · workflow
from zhivex_ai import WorkflowContextWorkflowContext(run_id: 'str', workflow_name: 'str', source: 'str', target: 'str', state: 'Mapping[str, JsonValue]', source_status: 'str', source_output: 'JsonValue | None' = None, resume_values: 'Mapping[str, JsonValue]' = <factory>) -> NoneWorkflowContext(run_id: 'str', workflow_name: 'str', source: 'str', target: 'str', state: 'Mapping[str, JsonValue]', source_status: 'str', source_output: 'JsonValue | None' = None, resume_values: 'Mapping[str, JsonValue]' = <factory>)WorkflowDefinitionMismatchError
Stable · errors
from zhivex_ai import WorkflowDefinitionMismatchErrorWorkflowDefinitionMismatchError (public type alias or constant; no callable signature)Raised when persisted state belongs to a different workflow definition.WorkflowEdge
Stable · workflow
from zhivex_ai import WorkflowEdgeWorkflowEdge(source: 'str', target: 'str', condition: 'WorkflowEdgeCondition | None' = None, name: 'str | None' = None, definition_revision: 'str | None' = None) -> NoneA directed workflow edge with optional application-owned identity.
Set ``definition_revision`` when the condition depends on configuration or
closure values that cannot be represented reliably by its Python source.WorkflowEdgeCondition
Stable · workflow
from zhivex_ai import WorkflowEdgeConditionWorkflowEdgeCondition(*args, **kwargs)WorkflowErrorPolicy
Stable · workflow
from zhivex_ai import WorkflowErrorPolicyWorkflowErrorPolicy(*args, **kwargs)WorkflowExecutionLease
Stable · workflow
from zhivex_ai import WorkflowExecutionLeaseWorkflowExecutionLease(run_id: 'str', owner_id: 'str', token: 'str', fencing_token: 'int', acquired_at_ms: 'int', renewed_at_ms: 'int', expires_at_ms: 'int') -> NoneWorkflowExecutionLease(run_id: 'str', owner_id: 'str', token: 'str', fencing_token: 'int', acquired_at_ms: 'int', renewed_at_ms: 'int', expires_at_ms: 'int')WorkflowFunctionContext
Stable · workflow
from zhivex_ai import WorkflowFunctionContextWorkflowFunctionContext(run_id: 'str', workflow_name: 'str', step_name: 'str', attempt: 'int', idempotency_key: 'str', input: 'JsonValue', state: 'Mapping[str, JsonValue]', resume_values: 'Mapping[str, JsonValue]' = <factory>, deps: 'Any' = None) -> NoneWorkflowFunctionContext(run_id: 'str', workflow_name: 'str', step_name: 'str', attempt: 'int', idempotency_key: 'str', input: 'JsonValue', state: 'Mapping[str, JsonValue]', resume_values: 'Mapping[str, JsonValue]' = <factory>, deps: 'Any' = None)WorkflowFunctionExecutor
Stable · workflow
from zhivex_ai import WorkflowFunctionExecutorWorkflowFunctionExecutor(*args, **kwargs)WorkflowFunctionResult
Stable · workflow
from zhivex_ai import WorkflowFunctionResultWorkflowFunctionResult(output: 'JsonValue' = None, state_patch: 'Mapping[str, JsonValue]' = <factory>, metadata: 'Mapping[str, JsonValue]' = <factory>) -> NoneWorkflowFunctionResult(output: 'JsonValue' = None, state_patch: 'Mapping[str, JsonValue]' = <factory>, metadata: 'Mapping[str, JsonValue]' = <factory>)WorkflowGraph
Stable · workflow
from zhivex_ai import WorkflowGraphWorkflowGraph(*, name: 'str', steps: 'Sequence[WorkflowStep]', edges: 'Sequence[WorkflowEdge]', definition_version: 'str' = '1', entrypoints: 'Sequence[str] | None' = None, checkpoint_store: 'WorkflowCheckpointStore | None' = None, run_store: 'AgentRunStore | None' = None, adapter: 'WorkflowAdapter | None' = None, interrupt_before: 'Mapping[str, str | None] | Sequence[str]' = (), interrupt_after: 'Mapping[str, str | None] | Sequence[str]' = (), max_concurrency: 'int | None' = None, lease_manager: 'WorkflowLeaseManager | None' = None, lease_ttl_ms: 'int' = 30000, lease_heartbeat_ms: 'int | None' = None, observer: 'AgentObserver | None' = None) -> 'None'Durable DAG workflow with append-only checkpoints and explicit resume.WorkflowInterrupt
Stable · workflow
from zhivex_ai import WorkflowInterruptWorkflowInterrupt(interrupt_id: 'str', node_name: 'str', reason: 'str | None' = None, payload: 'JsonValue | None' = None, created_at_ms: 'int | None' = None, phase: 'WorkflowInterruptPhase' = 'before', metadata: 'dict[str, JsonValue]' = <factory>) -> NoneWorkflowInterrupt(interrupt_id: 'str', node_name: 'str', reason: 'str | None' = None, payload: 'JsonValue | None' = None, created_at_ms: 'int | None' = None, phase: 'WorkflowInterruptPhase' = 'before', metadata: 'dict[str, JsonValue]' = <factory>)WorkflowInterruptError
Stable · errors
from zhivex_ai import WorkflowInterruptErrorWorkflowInterruptError (public type alias or constant; no callable signature)Raised when a workflow interrupt cannot be resumed safely.WorkflowInterruptPhase
Stable · workflow
from zhivex_ai import WorkflowInterruptPhaseWorkflowInterruptPhase(*args, **kwargs)WorkflowLeaseLostError
Stable · errors
from zhivex_ai import WorkflowLeaseLostErrorWorkflowLeaseLostError (public type alias or constant; no callable signature)Raised when a workflow worker no longer owns its execution lease.WorkflowLeaseManager
Stable · workflow
from zhivex_ai import WorkflowLeaseManagerWorkflowLeaseManager(*args, **kwargs)Base class for protocol classes.
Protocol classes are defined as::
class Proto(Protocol):
def meth(self) -> int:
...
Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).
For example::
class C:
def meth(self) -> int:
return 0
def func(x: Proto) -> int:
return x.meth()
func(C()) # Passes static type check
See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::
class GenProto[T](Protocol):
def meth(self) -> T:
...WorkflowNodeCheckpoint
Stable · workflow
from zhivex_ai import WorkflowNodeCheckpointWorkflowNodeCheckpoint(node_name: 'str', status: 'WorkflowNodeStatus' = 'pending', attempt: 'int' = 0, idempotency_key: 'str | None' = None, child_run_id: 'str | None' = None, output: 'JsonValue | None' = None, error: 'str | None' = None, started_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None, metadata: 'dict[str, JsonValue]' = <factory>, suspension: 'dict[str, JsonValue] | None' = None) -> NoneWorkflowNodeCheckpoint(node_name: 'str', status: 'WorkflowNodeStatus' = 'pending', attempt: 'int' = 0, idempotency_key: 'str | None' = None, child_run_id: 'str | None' = None, output: 'JsonValue | None' = None, error: 'str | None' = None, started_at_ms: 'int | None' = None, finished_at_ms: 'int | None' = None, metadata: 'dict[str, JsonValue]' = <factory>, suspension: 'dict[str, JsonValue] | None' = None)WorkflowNodeStatus
Stable · workflow
from zhivex_ai import WorkflowNodeStatusWorkflowNodeStatus(*args, **kwargs)WorkflowRetryPolicy
Stable · workflow
from zhivex_ai import WorkflowRetryPolicyWorkflowRetryPolicy(max_attempts: 'int' = 1, backoff_ms: 'int' = 250, max_backoff_ms: 'int' = 5000, retry_if: 'WorkflowRetryPredicate | None' = None) -> NoneRetry policy for a complete logical workflow step.
``WorkflowStep.max_retries`` remains the provider/model retry setting passed
to ``run_agent``. This policy is deliberately separate because retrying a
complete step can repeat tools or external side effects.WorkflowRetryPredicate
Stable · workflow
from zhivex_ai import WorkflowRetryPredicateWorkflowRetryPredicate(*args, **kwargs)WorkflowRunNotFoundError
Stable · errors
from zhivex_ai import WorkflowRunNotFoundErrorWorkflowRunNotFoundError (public type alias or constant; no callable signature)Raised when a requested workflow run or checkpoint cannot be found.WorkflowRunResult
Stable · workflow
from zhivex_ai import WorkflowRunResultWorkflowRunResult(run_id: 'str', name: 'str', session: 'AgentSession', state: 'WorkflowState', step_results: 'list[WorkflowStepResult]', text: 'str' = '', status: 'WorkflowRunStatus' = 'completed', trace: 'list[WorkflowTraceEvent]' = <factory>, state_snapshot: 'AgentRunState | None' = None, checkpoint: 'WorkflowCheckpoint | None' = None, forked_from_run_id: 'str | None' = None) -> NoneWorkflowRunResult(run_id: 'str', name: 'str', session: 'AgentSession', state: 'WorkflowState', step_results: 'list[WorkflowStepResult]', text: 'str' = '', status: 'WorkflowRunStatus' = 'completed', trace: 'list[WorkflowTraceEvent]' = <factory>, state_snapshot: 'AgentRunState | None' = None, checkpoint: 'WorkflowCheckpoint | None' = None, forked_from_run_id: 'str | None' = None)WorkflowRunStatus
Stable · workflow
from zhivex_ai import WorkflowRunStatusWorkflowRunStatus(*args, **kwargs)WorkflowState
Stable · workflow
from zhivex_ai import WorkflowStateWorkflowState(*args, **kwargs)WorkflowStep
Stable · workflow
from zhivex_ai import WorkflowStepWorkflowStep(name: 'str', agent: 'Agent | None' = None, prompt: 'str | None' = None, input_template: 'str | None' = None, output_key: 'str | None' = None, metadata_key: 'str | None' = None, max_retries: 'int | None' = None, timeout_ms: 'int | None' = None, error_policy: 'WorkflowErrorPolicy' = 'fail_fast', retry_policy: 'WorkflowRetryPolicy | None' = None, idempotency_key: 'str | None' = None, executor_ref: 'str | None' = None, metadata: 'dict[str, JsonValue]' = <factory>, executor: 'WorkflowFunctionExecutor | None' = None, definition_revision: 'str | None' = None) -> NoneA workflow node and its runtime configuration.
``definition_revision`` is an application-owned stable token for semantic
configuration that the SDK cannot inspect reliably, such as agent model or
instruction changes, tool configuration, and values captured by an
executor closure. Change it whenever that external configuration changes
so durable resume and fork operations fail closed on the new definition.WorkflowStepExecutor
Stable · workflow
from zhivex_ai import WorkflowStepExecutorWorkflowStepExecutor(*args, **kwargs)WorkflowStepExecutorRegistry
Stable · workflow
from zhivex_ai import WorkflowStepExecutorRegistryWorkflowStepExecutorRegistry() -> 'None'WorkflowStepOutcome
Stable · workflow
from zhivex_ai import WorkflowStepOutcomeWorkflowStepOutcome(workflow_run_id: 'str', node_id: 'str', activation_index: 'int', step_idempotency_key: 'str', status: 'WorkflowStepStatus', output: 'JsonValue' = None, state_patch: 'dict[str, JsonValue]' = <factory>, metadata: 'dict[str, JsonValue]' = <factory>, error: 'dict[str, JsonValue] | None' = None, suspension: 'dict[str, JsonValue] | None' = None, child_run_id: 'str | None' = None, schema_version: 'int' = 1) -> NoneWorkflowStepOutcome(workflow_run_id: 'str', node_id: 'str', activation_index: 'int', step_idempotency_key: 'str', status: 'WorkflowStepStatus', output: 'JsonValue' = None, state_patch: 'dict[str, JsonValue]' = <factory>, metadata: 'dict[str, JsonValue]' = <factory>, error: 'dict[str, JsonValue] | None' = None, suspension: 'dict[str, JsonValue] | None' = None, child_run_id: 'str | None' = None, schema_version: 'int' = 1)WorkflowStepRequest
Stable · workflow
from zhivex_ai import WorkflowStepRequestWorkflowStepRequest(workflow_name: 'str', definition_version: 'str', definition_digest: 'str', workflow_run_id: 'str', node_id: 'str', executor_ref: 'str', activation_index: 'int' = 0, attempt: 'int' = 1, state_revision: 'int' = 0, input: 'JsonValue' = None, state: 'dict[str, JsonValue]' = <factory>, metadata: 'dict[str, JsonValue]' = <factory>, checkpoint_id: 'str | None' = None, correlation_ids: 'dict[str, str]' = <factory>, schema_version: 'int' = 1) -> NoneWorkflowStepRequest(workflow_name: 'str', definition_version: 'str', definition_digest: 'str', workflow_run_id: 'str', node_id: 'str', executor_ref: 'str', activation_index: 'int' = 0, attempt: 'int' = 1, state_revision: 'int' = 0, input: 'JsonValue' = None, state: 'dict[str, JsonValue]' = <factory>, metadata: 'dict[str, JsonValue]' = <factory>, checkpoint_id: 'str | None' = None, correlation_ids: 'dict[str, str]' = <factory>, schema_version: 'int' = 1)WorkflowStepResult
Stable · workflow
from zhivex_ai import WorkflowStepResultWorkflowStepResult(name: 'str', status: 'WorkflowStepStatus', output: 'AgentRunResult | None' = None, error: 'Exception | None' = None, iteration: 'int | None' = None, output_text: 'str' = '', agent_run_id: 'str | None' = None, attempts: 'int' = 1) -> NoneWorkflowStepResult(name: 'str', status: 'WorkflowStepStatus', output: 'AgentRunResult | None' = None, error: 'Exception | None' = None, iteration: 'int | None' = None, output_text: 'str' = '', agent_run_id: 'str | None' = None, attempts: 'int' = 1)WorkflowStepStatus
Stable · workflow
from zhivex_ai import WorkflowStepStatusWorkflowStepStatus(*args, **kwargs)WorkflowStopCondition
Stable · workflow
from zhivex_ai import WorkflowStopConditionWorkflowStopCondition(*args, **kwargs)WorkflowTraceEvent
Stable · workflow
from zhivex_ai import WorkflowTraceEventWorkflowTraceEvent(type: 'str', workflow_name: 'str', step_name: 'str | None' = None, status: 'str | None' = None, iteration: 'int | None' = None, run_id: 'str | None' = None, error: 'str | None' = None) -> NoneWorkflowTraceEvent(type: 'str', workflow_name: 'str', step_name: 'str | None' = None, status: 'str | None' = None, iteration: 'int | None' = None, run_id: 'str | None' = None, error: 'str | None' = None)WorkflowTransition
Stable · workflow
from zhivex_ai import WorkflowTransitionWorkflowTransition(type: 'str', at_ms: 'int', node_name: 'str | None' = None, from_status: 'str | None' = None, to_status: 'str | None' = None, detail: 'dict[str, JsonValue]' = <factory>) -> NoneWorkflowTransition(type: 'str', at_ms: 'int', node_name: 'str | None' = None, from_status: 'str | None' = None, to_status: 'str | None' = None, detail: 'dict[str, JsonValue]' = <factory>)aclose_default_clients
Stable · transport
from zhivex_ai import aclose_default_clientsaclose_default_clients() -> 'None'Close the calling event loop's default HTTP pool before loop shutdown.agent_child_run_from_state
Beta · agent
from zhivex_ai import agent_child_run_from_stateagent_child_run_from_state(state: 'AgentRunState', *, tool_name: 'str | None' = None) -> 'AgentChildRun'agent_run_state_from_json
Stable · agent
from zhivex_ai import agent_run_state_from_jsonagent_run_state_from_json(value: 'str') -> 'AgentRunState'agent_run_state_to_json
Stable · agent
from zhivex_ai import agent_run_state_to_jsonagent_run_state_to_json(state: 'AgentRunState') -> 'str'allow_all_approval_policy
Beta · agent
from zhivex_ai import allow_all_approval_policyallow_all_approval_policy(request: 'ToolApprovalRequest') -> 'ApprovalDecision'anthropic_code_execution_tool
Beta · provider
from zhivex_ai import anthropic_code_execution_toolanthropic_code_execution_tool(*, name: 'str' = 'code_execution', tool_type: 'str' = 'code_execution_20260521', **extra: 'Any') -> 'HostedToolDefinition'anthropic_mcp_server
Beta · provider
from zhivex_ai import anthropic_mcp_serveranthropic_mcp_server(*, url: 'str', name: 'str', version: 'AnthropicMcpVersion' = 'legacy', authorization_token: 'str | None' = None, enabled: 'bool | None' = None, allowed_tools: 'list[str] | None' = None, tool_configuration: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'anthropic_web_fetch_tool
Beta · provider
from zhivex_ai import anthropic_web_fetch_toolanthropic_web_fetch_tool(*, name: 'str' = 'web_fetch', max_uses: 'int | None' = None, allowed_domains: 'list[str] | None' = None, blocked_domains: 'list[str] | None' = None, citations_enabled: 'bool | None' = None, max_content_tokens: 'int | None' = None, use_cache: 'bool | None' = None, response_inclusion: "Literal['full', 'excluded'] | None" = None, tool_type: 'str' = 'web_fetch_20260318', **extra: 'Any') -> 'HostedToolDefinition'anthropic_web_search_tool
Beta · provider
from zhivex_ai import anthropic_web_search_toolanthropic_web_search_tool(*, name: 'str' = 'web_search', max_uses: 'int | None' = None, allowed_domains: 'list[str] | None' = None, blocked_domains: 'list[str] | None' = None, user_location: 'dict[str, Any] | None' = None, tool_type: 'str' = 'web_search_20260318', **extra: 'Any') -> 'HostedToolDefinition'apply_safety_policy_to_agent
Beta · safety
from zhivex_ai import apply_safety_policy_to_agentapply_safety_policy_to_agent(agent: 'Agent', policy: 'SafetyPolicy') -> 'Agent'assistant
Beta · messages
from zhivex_ai import assistantassistant(input: 'str | list[ContentPart]') -> 'ModelMessage'azure_openai_computer_use_tool
Beta · provider
from zhivex_ai import azure_openai_computer_use_toolazure_openai_computer_use_tool(*, environment: 'str | None' = None, display_width: 'int | None' = None, display_height: 'int | None' = None, tool_type: 'str' = 'computer_use_preview', **extra: 'object') -> 'HostedToolDefinition'azure_openai_file_search_tool
Beta · provider
from zhivex_ai import azure_openai_file_search_toolazure_openai_file_search_tool(*, vector_store_ids: 'list[str]', filters: 'dict[str, object] | None' = None, max_num_results: 'int | None' = None, **extra: 'object') -> 'HostedToolDefinition'azure_openai_mcp_approval_response
Beta · provider
from zhivex_ai import azure_openai_mcp_approval_responseazure_openai_mcp_approval_response(*, approval_request_id: 'str', approve: 'bool', id: 'str | None' = None, reason: 'str | None' = None) -> 'ProviderDataPart'azure_openai_mcp_tool
Beta · provider
from zhivex_ai import azure_openai_mcp_toolazure_openai_mcp_tool(*, server_url: 'str | None' = None, server_label: 'str | None' = None, headers: 'dict[str, str] | None' = None, allowed_tools: 'list[str] | None' = None, require_approval: 'str | None' = None, **extra: 'object') -> 'HostedToolDefinition'azure_openai_response_reference
Beta · provider
from zhivex_ai import azure_openai_response_referenceazure_openai_response_reference(*, response_id: 'str') -> 'ProviderDataPart'azure_openai_web_search_tool
Beta · provider
from zhivex_ai import azure_openai_web_search_toolazure_openai_web_search_tool(*, search_context_size: 'str | None' = None, user_location: 'dict[str, object] | None' = None, tool_type: 'str' = 'web_search_preview', **extra: 'object') -> 'HostedToolDefinition'build_provider_support_rows
Beta · provider-support
from zhivex_ai import build_provider_support_rowsbuild_provider_support_rows(providers: 'Mapping[str, ProviderBundle] | Iterable[ProviderBundle]', *, validated_release_certifications: 'Iterable[str]' = ()) -> 'list[ProviderSupportRow]'Build support rows without inferring live evidence from provider tier.
``validated_release_certifications`` is an explicit trust boundary. Callers
must populate it only with provider names produced by a separate release
evidence validator for the exact artifact being described. An empty value
is intentionally fail-closed, including for Tier-1 providers.cancel_agent_run
Stable · agent
from zhivex_ai import cancel_agent_runcancel_agent_run(store: 'AgentRunStore', run_id: 'str', *, reason: 'str | None' = None, now_ms: 'int | None' = None, cancellation_token: 'Any' = None) -> 'AgentRunState | None'cancel_agent_run_tree
Stable · agent
from zhivex_ai import cancel_agent_run_treecancel_agent_run_tree(store: 'AgentRunStore', run_id: 'str', *, reason: 'str | None' = None, now_ms: 'int | None' = None, cancellation_token: 'Any' = None) -> 'AgentRunTreeCancellationResult'cancel_workflow
Stable · workflow
from zhivex_ai import cancel_workflowcancel_workflow(workflow: 'WorkflowGraph', run_id: 'str', *, reason: 'str | None' = None, session: 'AgentSession | None' = None) -> 'WorkflowRunResult'clear_agent_session_skills
Stable · agent
from zhivex_ai import clear_agent_session_skillsclear_agent_session_skills(session: 'AgentSession') -> 'AgentSession'collect_ui_message
Beta · ui
from zhivex_ai import collect_ui_messagecollect_ui_message(result: 'Any', message_id: 'str | None' = None) -> 'UIMessage'create_a2a_agent_card
Beta · protocol
from zhivex_ai import create_a2a_agent_cardcreate_a2a_agent_card(agent: 'Agent', *, url: 'str', version: 'str', description: 'str | None' = None, skills: 'list[A2AAgentSkill] | None' = None, provider: 'dict[str, str] | None' = None, documentation_url: 'str | None' = None, security_schemes: 'dict[str, JsonValue] | None' = None, security: 'list[dict[str, list[str]]] | None' = None) -> 'A2AAgentCard'create_a2a_app
Beta · protocol
from zhivex_ai import create_a2a_appcreate_a2a_app(*, executor: 'A2AAgentExecutor', card: 'A2AAgentCard', authorize: 'Callable[[Any], bool | Awaitable[bool]] | None' = None, max_request_bytes: 'int | None' = None, limits: 'ProtocolLimits | None' = None, task_store: 'Any' = None, request_context_builder: 'Any' = None, queue_manager: 'Any' = None)Create an A2A v1 server using the official Python SDK.
The app exposes both HTTP+JSON under ``/a2a`` and JSON-RPC at
``/a2a/rpc``. The official task store and request handler own the wire
protocol, task lifecycle, version headers, and streaming envelopes.create_agent_evaluation_dataset_from_traces
Beta · agent
from zhivex_ai import create_agent_evaluation_dataset_from_tracescreate_agent_evaluation_dataset_from_traces(traces: 'list[AgentRunState]', *, prompt_extractor: 'AgentEvaluationTracePromptExtractor', expectations_extractor: 'AgentEvaluationTraceExpectationsExtractor', name_extractor: 'AgentEvaluationTraceNameExtractor | None' = None, metadata_extractor: 'AgentEvaluationTraceMetadataExtractor | None' = None) -> 'list[AgentEvaluationCase]'Build evaluation cases without implicitly copying persisted prompt or output data.
Both content-bearing fields are supplied by application-owned extractors so
applications can apply their own consent, retention, and redaction policy.create_agent_evaluation_fixture
Beta · agent
from zhivex_ai import create_agent_evaluation_fixturecreate_agent_evaluation_fixture(*, name: 'str', dataset: 'list[AgentEvaluationCase]', expected_ok: 'bool' = True, metadata: 'dict[str, JsonValue] | None' = None) -> 'AgentEvaluationFixture'create_agent_evaluation_report
Beta · agent
from zhivex_ai import create_agent_evaluation_reportcreate_agent_evaluation_report(result: 'AgentEvaluationResult', *, metadata: 'dict[str, JsonValue] | None' = None) -> 'AgentEvaluationReport'create_agent_evaluation_trajectory
Beta · agent
from zhivex_ai import create_agent_evaluation_trajectorycreate_agent_evaluation_trajectory(result: 'AgentRunResult') -> 'AgentEvaluationTrajectory | None'Create a trace projection that excludes messages, tool payloads, and error bodies.create_agent_playground_app
Beta · protocol
from zhivex_ai import create_agent_playground_appcreate_agent_playground_app(*, agents: 'AgentResolver', authorize: 'Callable[[Any], bool | Awaitable[bool]] | None' = None, max_request_bytes: 'int | None' = None, limits: 'ProtocolLimits | None' = None, run_options_resolver: 'ProtocolRunOptionsResolver | None' = None, error_mapper: 'ProtocolErrorMapper | None' = None, on_protocol_event: 'ProtocolEventCallback | None' = None, event_store: 'ResponsesEventStore | None' = None)Create a local playground plus the Responses-compatible endpoint.create_agent_run_snapshot
Stable · agent
from zhivex_ai import create_agent_run_snapshotcreate_agent_run_snapshot(state: 'AgentRunState') -> 'AgentRunSnapshot'create_agent_run_tree_snapshot
Beta · observability
from zhivex_ai import create_agent_run_tree_snapshotcreate_agent_run_tree_snapshot(store: 'AgentRunStore', run_id: 'str') -> 'AgentRunTreeSnapshot'create_agent_session
Stable · agent
from zhivex_ai import create_agent_sessioncreate_agent_session(*, id: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, summary: 'str | None' = None, state: 'dict[str, JsonValue] | None' = None, metadata: 'dict[str, Any] | None' = None) -> 'AgentSession'create_agent_trace_artifact
Beta · observability
from zhivex_ai import create_agent_trace_artifactcreate_agent_trace_artifact(state: 'AgentRunState', *, include_messages: 'bool' = False, include_tool_inputs: 'bool' = False, output_preview_length: 'int' = 500) -> 'AgentTraceArtifact'create_agent_trace_collector
Beta · observability
from zhivex_ai import create_agent_trace_collectorcreate_agent_trace_collector() -> 'AgentTraceCollector'create_anthropic
Stable · provider
from zhivex_ai import create_anthropiccreate_anthropic(*, api_key: 'str | None' = None, base_url: 'str' = 'https://api.anthropic.com/v1', anthropic_version: 'str' = '2023-06-01', beta_headers: 'str | list[str] | tuple[str, ...] | None' = None, fetch: 'Fetcher | None' = None) -> 'ProviderBundle'create_approval_policy
Beta · safety
from zhivex_ai import create_approval_policycreate_approval_policy(*, preset: 'ApprovalPolicyPreset' = 'review_sensitive', sensitive_tool_names: 'list[str] | None' = None, allow_tool_names: 'list[str] | None' = None, deny_tool_names: 'list[str] | None' = None) -> 'ApprovalPolicy'create_azure_openai
Stable · provider
from zhivex_ai import create_azure_openaicreate_azure_openai(*, api_key: 'str | None' = None, endpoint: 'str | None' = None, api_version: 'str' = '2024-10-21', entra_token: 'str | None' = None, entra_token_provider: 'AzureOpenAITokenProvider | None' = None, fetch: 'Fetcher | None' = None, realtime_url: 'str | None' = None, browser_token_url: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None) -> 'ProviderBundle'create_bedrock
Experimental · provider
from zhivex_ai import create_bedrockcreate_bedrock(*, client: 'BedrockClient | None' = None, region: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None)native-only provider
create_budget_guard
Beta · safety
from zhivex_ai import create_budget_guardcreate_budget_guard(*, max_steps: 'int | None' = None, max_tool_calls: 'int | None' = None, max_tool_errors: 'int | None' = None, max_input_tokens: 'int | None' = None, max_output_tokens: 'int | None' = None, max_total_tokens: 'int | None' = None, include_child_runs: 'bool' = True) -> 'BudgetGuard'create_cached_generate_middleware
Beta · middleware
from zhivex_ai import create_cached_generate_middlewarecreate_cached_generate_middleware(*, cache: 'GenerateCache', get_key: 'Callable[[ModelGenerateInput, LanguageModel], str] | None' = None) -> 'GenerateMiddleware'create_circuit_breaker_middleware
Beta · middleware
from zhivex_ai import create_circuit_breaker_middlewarecreate_circuit_breaker_middleware(*, failure_threshold: 'int' = 3, cooldown_ms: 'int' = 30000, is_failure: 'Callable[[Exception], bool] | None' = None, on_state_change: 'Callable[[dict[str, Any]], Awaitable[None] | None] | None' = None) -> 'GenerateMiddleware'create_dbos_workflow_adapter
Beta · workflow
from zhivex_ai import create_dbos_workflow_adaptercreate_dbos_workflow_adapter(callback: 'WorkflowStepExecutor') -> 'CallbackWorkflowAdapter'beta named-engine factory; not a certified DBOS integration
create_deepseek
Stable · provider
from zhivex_ai import create_deepseekcreate_deepseek(*, api_key: 'str | None' = None, base_url: 'str | None' = None, fetch: 'Fetcher | None' = None) -> 'ProviderBundle'tier-1 provider; native provider options remain beta
create_file_generate_cache
Beta · middleware
from zhivex_ai import create_file_generate_cachecreate_file_generate_cache(*, dir: 'str') -> 'FileGenerateCache'create_gateway
Stable · gateway
from zhivex_ai import create_gatewaycreate_gateway(config: 'GatewayConfig')create_gemini
Stable · provider
from zhivex_ai import create_geminicreate_gemini(*, api_key: 'str | None' = None, base_url: 'str' = 'https://generativelanguage.googleapis.com/v1beta', fetch: 'Fetcher | None' = None, realtime_url: 'str | None' = None, auth_token_url: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None) -> 'ProviderBundle'create_hierarchical_agent_trace
Beta · observability
from zhivex_ai import create_hierarchical_agent_tracecreate_hierarchical_agent_trace(store: 'AgentRunStore', run_id: 'str') -> 'HierarchicalAgentTrace'create_in_memory_agent_memory_store
Stable · agent
from zhivex_ai import create_in_memory_agent_memory_storecreate_in_memory_agent_memory_store(*, summary_config: 'SummaryConfig | None' = None) -> 'InMemoryAgentMemory'create_in_memory_agent_run_store
Stable · agent
from zhivex_ai import create_in_memory_agent_run_storecreate_in_memory_agent_run_store() -> 'InMemoryAgentRunStore'create_in_memory_checkpoint_store
Stable · agent
from zhivex_ai import create_in_memory_checkpoint_storecreate_in_memory_checkpoint_store() -> 'InMemoryAgentCheckpointStore'create_in_memory_generate_cache
Beta · middleware
from zhivex_ai import create_in_memory_generate_cachecreate_in_memory_generate_cache() -> 'InMemoryGenerateCache'create_in_memory_workflow_checkpoint_store
Stable · workflow
from zhivex_ai import create_in_memory_workflow_checkpoint_storecreate_in_memory_workflow_checkpoint_store() -> 'InMemoryWorkflowCheckpointStore'create_in_memory_workflow_lease_manager
Stable · workflow
from zhivex_ai import create_in_memory_workflow_lease_managercreate_in_memory_workflow_lease_manager() -> 'InMemoryWorkflowLeaseManager'create_kimi
Stable · provider
from zhivex_ai import create_kimicreate_kimi(*, api_key: 'str | None' = None, base_url: 'str | None' = None, fetch: 'Fetcher | None' = None) -> 'ProviderBundle'tier-1 provider; native extras remain beta
create_mcp_tool_registry
Stable · agent
from zhivex_ai import create_mcp_tool_registrycreate_mcp_tool_registry(server: 'MCPServerConfig', *, prefix: 'str | None' = None, include: 'Iterable[str] | None' = None, exclude: 'Iterable[str] | None' = None, trusted_tools: 'Iterable[str] | None' = None, name_transform: "Literal['preserve', 'snake_case']" = 'snake_case') -> 'ToolRegistry'create_meta
Stable · provider
from zhivex_ai import create_metacreate_meta(*, api_key: 'str | None' = None, base_url: 'str' = 'https://api.meta.ai/v1', fetch: 'Fetcher | None' = None) -> 'ProviderBundle'tier-1 provider for the portable Muse Spark contract; native extras remain beta
create_mock_language_model
Beta · agent
from zhivex_ai import create_mock_language_modelcreate_mock_language_model(*, provider: 'str' = 'mock', model_id: 'str' = 'mock-model', responses: 'list[GenerateResult] | None' = None, stream_events: 'list[list[StreamEvent]] | None' = None) -> 'LanguageModel'create_mock_tool
Beta · agent
from zhivex_ai import create_mock_toolcreate_mock_tool(name: 'str', *, outputs: 'list[JsonValue] | None' = None, errors: 'list[str | Exception] | None' = None) -> 'ToolDefinition'create_model_catalog
Stable · catalog
from zhivex_ai import create_model_catalogcreate_model_catalog(entries: 'Iterable[ModelCatalogEntry]') -> 'ModelCatalog'create_ollama
Experimental · provider
from zhivex_ai import create_ollamacreate_ollama(*, api_key: 'str | None' = 'ollama', base_url: 'str' = 'http://localhost:11434/v1', fetch: 'Fetcher | None' = None)compatibility provider
create_openai
Stable · provider
from zhivex_ai import create_openaicreate_openai(*, api_key: 'str | None' = None, base_url: 'str' = 'https://api.openai.com/v1', fetch: 'Fetcher | None' = None, realtime_url: 'str | None' = None, browser_token_url: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None) -> 'ProviderBundle'create_openrouter
Experimental · provider
from zhivex_ai import create_openroutercreate_openrouter(*, api_key: 'str | None' = None, base_url: 'str' = 'https://openrouter.ai/api/v1', fetch: 'Fetcher | None' = None)native-only provider
create_otel_agent_observer
Beta · observability
from zhivex_ai import create_otel_agent_observercreate_otel_agent_observer(*, tracer_name: 'str' = 'zhivex_ai.agent', version: 'str | None' = None) -> 'OTelAgentObserver'create_postgres_agent_memory_store
Stable · agent
from zhivex_ai import create_postgres_agent_memory_storecreate_postgres_agent_memory_store(dsn: 'str', *, summary_config: 'SummaryConfig | None' = None, table_prefix: 'str' = 'zhivex_ai') -> 'PostgresAgentMemoryStore'create_postgres_agent_run_store
Stable · agent
from zhivex_ai import create_postgres_agent_run_storecreate_postgres_agent_run_store(dsn: 'str', *, table_prefix: 'str' = 'zhivex_agent') -> 'PostgresAgentRunStore'create_postgres_checkpoint_store
Stable · agent
from zhivex_ai import create_postgres_checkpoint_storecreate_postgres_checkpoint_store(dsn: 'str', *, table_prefix: 'str' = 'zhivex_ai') -> 'PostgresAgentCheckpointStore'create_postgres_workflow_checkpoint_store
Stable · workflow
from zhivex_ai import create_postgres_workflow_checkpoint_storecreate_postgres_workflow_checkpoint_store(dsn: 'str | None' = None, *, table_prefix: 'str' = 'zhivex_ai', namespace: 'str' = 'default', pool: 'Any | None' = None, pool_min_size: 'int' = 1, pool_max_size: 'int' = 5) -> 'PostgresWorkflowCheckpointStore'create_postgres_workflow_lease_manager
Stable · workflow
from zhivex_ai import create_postgres_workflow_lease_managercreate_postgres_workflow_lease_manager(dsn: 'str | None' = None, *, table_prefix: 'str' = 'zhivex_ai', namespace: 'str' = 'default', pool: 'Any | None' = None, pool_min_size: 'int' = 1, pool_max_size: 'int' = 5) -> 'PostgresWorkflowLeaseManager'create_prefect_workflow_adapter
Beta · workflow
from zhivex_ai import create_prefect_workflow_adaptercreate_prefect_workflow_adapter(callback: 'WorkflowStepExecutor') -> 'CallbackWorkflowAdapter'beta named-engine factory; not a certified Prefect integration
create_qwen
Stable · provider
from zhivex_ai import create_qwencreate_qwen(*, api_key: 'str | None' = None, region: 'QwenRegion' = 'intl', base_url: 'str | None' = None, responses_base_url: 'str | None' = None, fetch: 'Fetcher | None' = None) -> 'ProviderBundle'tier-1 provider; native extras remain beta
create_redaction_policy
Beta · safety
from zhivex_ai import create_redaction_policycreate_redaction_policy(*, rules: 'list[RedactionRule] | None' = None, include_emails: 'bool' = False, replacement: 'str' = '[REDACTED]') -> 'RedactionPolicy'create_responses_app
Beta · protocol
from zhivex_ai import create_responses_appcreate_responses_app(*, agents: 'AgentResolver', authorize: 'Callable[[Any], bool | Awaitable[bool]] | None' = None, max_request_bytes: 'int | None' = None, limits: 'ProtocolLimits | None' = None, run_options_resolver: 'ProtocolRunOptionsResolver | None' = None, error_mapper: 'ProtocolErrorMapper | None' = None, on_protocol_event: 'ProtocolEventCallback | None' = None, event_store: 'ResponsesEventStore | None' = None)Create an optional FastAPI app with ``POST /v1/responses``.create_restate_workflow_adapter
Beta · workflow
from zhivex_ai import create_restate_workflow_adaptercreate_restate_workflow_adapter(callback: 'WorkflowStepExecutor') -> 'CallbackWorkflowAdapter'beta named-engine factory; not a certified Restate integration
create_safety_policy
Beta · safety
from zhivex_ai import create_safety_policycreate_safety_policy(*, preset: 'SafetyPolicyPreset' = 'review_sensitive', approval: 'ApprovalPolicy | ApprovalPolicyOptions | bool | None' = None, redaction: 'RedactionPolicy | bool | None' = None, budget: 'BudgetGuard | bool | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, input_guardrails: 'list[InputGuardrail] | None' = None, output_guardrails: 'list[OutputGuardrail] | None' = None) -> 'SafetyPolicy'create_sqlite_agent_memory_store
Stable · agent
from zhivex_ai import create_sqlite_agent_memory_storecreate_sqlite_agent_memory_store(path: 'str', *, summary_config: 'SummaryConfig | None' = None, namespace: 'str' = 'default') -> 'SQLiteAgentMemoryStore'create_sqlite_agent_run_store
Stable · agent
from zhivex_ai import create_sqlite_agent_run_storecreate_sqlite_agent_run_store(path: 'str', *, namespace: 'str' = 'default') -> 'SQLiteAgentRunStore'create_sqlite_checkpoint_store
Stable · agent
from zhivex_ai import create_sqlite_checkpoint_storecreate_sqlite_checkpoint_store(path: 'str', *, namespace: 'str' = 'default') -> 'SQLiteAgentCheckpointStore'create_sqlite_workflow_checkpoint_store
Stable · workflow
from zhivex_ai import create_sqlite_workflow_checkpoint_storecreate_sqlite_workflow_checkpoint_store(path: 'str', *, namespace: 'str' = 'default') -> 'SQLiteWorkflowCheckpointStore'create_sqlite_workflow_lease_manager
Stable · workflow
from zhivex_ai import create_sqlite_workflow_lease_managercreate_sqlite_workflow_lease_manager(path: 'str', *, namespace: 'str' = 'default') -> 'SQLiteWorkflowLeaseManager'create_subagent_tool
Beta · agent
from zhivex_ai import create_subagent_toolcreate_subagent_tool(*, name: 'str', agent: 'Agent', parent_run_id: 'str | None' = None, description: 'str | None' = None, runtime: 'AgentRuntime | None' = None, hooks: 'Iterable[AgentHooks] | None' = None) -> 'ToolDefinition'create_telemetry_middleware
Beta · middleware
from zhivex_ai import create_telemetry_middlewarecreate_telemetry_middleware(*, on_event: 'Callable[[dict[str, Any]], Awaitable[None] | None]') -> 'GenerateMiddleware'create_temporal_workflow_adapter
Beta · workflow
from zhivex_ai import create_temporal_workflow_adaptercreate_temporal_workflow_adapter(callback: 'WorkflowStepExecutor') -> 'CallbackWorkflowAdapter'beta named-engine factory; not a certified Temporal integration
create_text_message
Beta · messages
from zhivex_ai import create_text_messagecreate_text_message(role: 'MessageRole', text: 'str') -> 'ModelMessage'create_ui_message_json_response
Beta · transport
from zhivex_ai import create_ui_message_json_responsecreate_ui_message_json_response(messages: 'list[UIMessage]', *, status_code: 'int' = 200, headers: 'Mapping[str, str] | None' = None) -> 'HTTPResponse'create_ui_message_lines_response
Beta · transport
from zhivex_ai import create_ui_message_lines_responsecreate_ui_message_lines_response(messages: 'list[UIMessage]', *, status_code: 'int' = 200, headers: 'Mapping[str, str] | None' = None) -> 'HTTPResponse'create_vertex
Stable · provider
from zhivex_ai import create_vertexcreate_vertex(*, access_token: 'str | None' = None, project_id: 'str | None' = None, location: 'str' = 'us-central1', api_version: 'str' = 'v1', base_url: 'str | None' = None, fetch: 'Fetcher | None' = None, realtime_url: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None) -> 'ProviderBundle'create_vllm
Stable · provider
from zhivex_ai import create_vllmcreate_vllm(*, api_key: 'str | None' = None, base_url: 'str | None' = None, fetch: 'Fetcher | None' = None, realtime_url: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None) -> 'ProviderBundle'tier-1 provider; some capabilities are model/task dependent
default_model_catalog
Beta · catalog
from zhivex_ai import default_model_catalogdefault_model_catalog (public type alias or constant; no callable signature)deny_all_approval_policy
Beta · agent
from zhivex_ai import deny_all_approval_policydeny_all_approval_policy(request: 'ToolApprovalRequest') -> 'ApprovalDecision'deserialize_agent_run_state
Stable · agent
from zhivex_ai import deserialize_agent_run_statedeserialize_agent_run_state(payload: 'dict[str, Any]') -> 'AgentRunState'deserialize_ui_message
Beta · ui
from zhivex_ai import deserialize_ui_messagedeserialize_ui_message(value: 'str') -> 'UIMessage'deserialize_ui_message_chunk
Beta · ui
from zhivex_ai import deserialize_ui_message_chunkdeserialize_ui_message_chunk(value: 'str') -> 'UIMessageChunk'deserialize_workflow_checkpoint
Stable · workflow
from zhivex_ai import deserialize_workflow_checkpointdeserialize_workflow_checkpoint(payload: 'dict[str, Any]') -> 'WorkflowCheckpoint'discover_mcp_tools
Stable · agent
from zhivex_ai import discover_mcp_toolsdiscover_mcp_tools(server: 'MCPServerConfig', *, prefix: 'str | None' = None, include: 'Iterable[str] | None' = None, exclude: 'Iterable[str] | None' = None, trusted_tools: 'Iterable[str] | None' = None) -> 'ToolSet'discover_skills
Stable · skills
from zhivex_ai import discover_skillsdiscover_skills(*, cwd: 'str | Path | None' = None, search_up: 'bool' = True, extra_paths: 'list[str | Path] | None' = None) -> 'SkillSet'embed
Stable · foundation
from zhivex_ai import embedembed(*, model: 'EmbeddingModel', value: 'str | list[str]', timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'EmbedOutput'embed_content
Stable · foundation
from zhivex_ai import embed_contentembed_content(*, model: 'EmbeddingModel', value: 'EmbeddingContent', timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'EmbedOutput'embed_content_many
Stable · foundation
from zhivex_ai import embed_content_manyembed_content_many(*, model: 'EmbeddingModel', values: 'list[EmbeddingContent]', timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'EmbedOutput'embed_many
Stable · foundation
from zhivex_ai import embed_manyembed_many(*, model: 'EmbeddingModel', values: 'list[str]', timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'EmbedOutput'estimate_agent_run_cost
Beta · observability
from zhivex_ai import estimate_agent_run_costestimate_agent_run_cost(state: 'AgentRunState', pricing: 'TokenPricing | ModelCatalog') -> 'CostEstimate'estimate_token_cost
Beta · observability
from zhivex_ai import estimate_token_costestimate_token_cost(usage: 'TokenUsage | None', pricing: 'TokenPricing') -> 'CostEstimate'fail_agent_run_resume_claim
Beta · agent
from zhivex_ai import fail_agent_run_resume_claimfail_agent_run_resume_claim(store: 'AgentRunStore', run_id: 'str', *, claim_token: 'str', reason: 'str', now_ms: 'int | None' = None) -> 'AgentRunState | None'Atomically fail one known approval-resume claim without retrying its tool.fork_workflow
Stable · workflow
from zhivex_ai import fork_workflowfork_workflow(workflow: 'WorkflowGraph', run_id: 'str', *, checkpoint_id: 'str | None' = None, state_updates: 'Mapping[str, JsonValue] | None' = None, idempotency_key: 'str | None' = None, deps: 'Any' = None, session: 'AgentSession | None' = None) -> 'WorkflowRunResult'from_ui_message
Beta · ui
from zhivex_ai import from_ui_messagefrom_ui_message(message: 'UIMessage') -> 'ModelMessage'from_ui_messages
Beta · ui
from zhivex_ai import from_ui_messagesfrom_ui_messages(messages: 'list[UIMessage]') -> 'list[ModelMessage]'gemini_code_execution_tool
Beta · provider
from zhivex_ai import gemini_code_execution_toolgemini_code_execution_tool(**config: 'Any') -> 'HostedToolDefinition'gemini_computer_use_tool
Beta · provider
from zhivex_ai import gemini_computer_use_toolgemini_computer_use_tool(**config: 'Any') -> 'HostedToolDefinition'gemini_file_search_tool
Beta · provider
from zhivex_ai import gemini_file_search_toolgemini_file_search_tool(*, file_search_store_names: 'list[str]', filters: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'gemini_google_maps_tool
Beta · provider
from zhivex_ai import gemini_google_maps_toolgemini_google_maps_tool(**config: 'Any') -> 'HostedToolDefinition'gemini_google_search_tool
Beta · provider
from zhivex_ai import gemini_google_search_toolgemini_google_search_tool(*, exclude_domains: 'list[str] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'gemini_hosted_tool
Beta · provider
from zhivex_ai import gemini_hosted_toolgemini_hosted_tool(tool_type: 'str', /, *, name: 'str | None' = None, tool_class: 'str | None' = None, **config: 'Any') -> 'HostedToolDefinition'gemini_url_context_tool
Beta · provider
from zhivex_ai import gemini_url_context_toolgemini_url_context_tool(**config: 'Any') -> 'HostedToolDefinition'generate_grounded_text
Stable · foundation
from zhivex_ai import generate_grounded_textgenerate_grounded_text(*, model: 'GroundedLanguageModel', prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, system: 'str | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, config: 'PortableGroundingConfig | None' = None, retrieval: 'PortableRetrievalConfig | None' = None, provider_options: 'dict[str, object] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'GenerateGroundedTextOutput'generate_object
Stable · foundation
from zhivex_ai import generate_objectgenerate_object(*, model: 'Any', schema: 'Any', prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, system: 'str | None' = None, mode: 'str' = 'auto', schema_name: 'str | None' = None, schema_description: 'str | None' = None, **kwargs: 'Any') -> 'GenerateObjectOutput'generate_speech
Beta · foundation
from zhivex_ai import generate_speechgenerate_speech(*, model: 'SpeechModel', input: 'str', voice: 'str | None' = None, config: 'PortableSpeechConfig | None' = None, provider_options: 'dict[str, object] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'SpeechOutput'generate_text
Stable · foundation
from zhivex_ai import generate_textgenerate_text(*, model: 'LanguageModel', prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, system: 'str | None' = None, tools: 'ToolSet | None' = None, tool_choice: 'ToolChoice | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, max_steps: 'int | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, retrieval: 'PortableRetrievalConfig | None' = None, structured_output: 'Any' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'GenerateTextOutput'get_agent_capabilities
Beta · messages
from zhivex_ai import get_agent_capabilitiesget_agent_capabilities(model: 'LanguageModel') -> 'AgentCapabilities'get_agent_session_skills
Stable · agent
from zhivex_ai import get_agent_session_skillsget_agent_session_skills(session: 'AgentSession') -> 'list[str]'get_agent_support_tier
Beta · messages
from zhivex_ai import get_agent_support_tierget_agent_support_tier(model: 'LanguageModel') -> 'AgentSupportTier'get_azure_openai_response_id
Beta · provider
from zhivex_ai import get_azure_openai_response_idget_azure_openai_response_id(value: 'Any') -> 'str | None'get_azure_openai_response_reference
Beta · provider
from zhivex_ai import get_azure_openai_response_referenceget_azure_openai_response_reference(value: 'Any') -> 'AzureOpenAIResponseReference | None'get_hosted_tool_class
Beta · messages
from zhivex_ai import get_hosted_tool_classget_hosted_tool_class(tool_definition: 'HostedToolDefinition') -> 'HostedToolClass'get_last_provider_data_part
Beta · messages
from zhivex_ai import get_last_provider_data_partget_last_provider_data_part(value: 'Any', *, provider: 'str | None' = None) -> 'ProviderDataPart | None'get_openai_response_id
Beta · provider
from zhivex_ai import get_openai_response_idget_openai_response_id(value: 'Any') -> 'str | None'get_openai_response_reference
Beta · provider
from zhivex_ai import get_openai_response_referenceget_openai_response_reference(value: 'Any') -> 'OpenAIResponseReference | None'get_pending_agent_approvals
Stable · agent
from zhivex_ai import get_pending_agent_approvalsget_pending_agent_approvals(store: 'AgentRunStore', run_id: 'str') -> 'list[PendingApproval]'get_provider_data_parts
Beta · messages
from zhivex_ai import get_provider_data_partsget_provider_data_parts(value: 'Any', *, provider: 'str | None' = None) -> 'list[ProviderDataPart]'handoff_to
Stable · agent
from zhivex_ai import handoff_tohandoff_to(target_agent: 'str', *, input: 'str | None' = None, metadata: 'dict[str, Any] | None' = None) -> 'dict[str, Any]'hosted_tool
Beta · messages
from zhivex_ai import hosted_toolhosted_tool(*, name: 'str', type: 'str', provider: 'str | None' = None, config: 'Any' = None, tool_class: 'HostedToolClass | None' = None, requires_approval: 'bool | None' = None, metadata: 'dict[str, Any] | None' = None) -> 'HostedToolDefinition'install_skill
Beta · skills
from zhivex_ai import install_skillinstall_skill(source: 'str | Path', *, project_root: 'str | Path | None' = None, lock: 'bool' = True, registry_url: 'str | None' = None, trust_remote_code: 'bool' = False) -> 'InstalledSkill'is_callable_tool_definition
Beta · messages
from zhivex_ai import is_callable_tool_definitionis_callable_tool_definition(tool_definition: 'ToolDefinition | HostedToolDefinition') -> 'TypeGuard[ToolDefinition]'is_hosted_tool_class
Beta · messages
from zhivex_ai import is_hosted_tool_classis_hosted_tool_class(tool_definition: 'HostedToolDefinition', tool_class: 'HostedToolClass') -> 'bool'is_hosted_tool_definition
Beta · messages
from zhivex_ai import is_hosted_tool_definitionis_hosted_tool_definition(tool_definition: 'ToolDefinition | HostedToolDefinition') -> 'TypeGuard[HostedToolDefinition]'judge_agent_evaluation
Beta · agent
from zhivex_ai import judge_agent_evaluationjudge_agent_evaluation(result: 'AgentEvaluationResult', judge: 'Callable[[AgentEvaluationResult], AgentEvaluationJudgeResult | Awaitable[AgentEvaluationJudgeResult]] | None' = None) -> 'AgentEvaluationJudgeResult'Score an evaluation with a custom judge or deterministic expectations.
The built-in path does not call a language model. It reports the case pass
rate produced by :class:`AgentEvaluationExpectations`. Applications may
supply a provider-agnostic callable for rubric or model-based judging.kimi_formula_toolset
Beta · provider
from zhivex_ai import kimi_formula_toolsetkimi_formula_toolset(client: 'KimiFormulaClient', formula_uris: 'list[str] | tuple[str, ...]', options: 'RetryOptions | None' = None) -> 'dict[str, ToolDefinition]'list_installed_skills
Beta · skills
from zhivex_ai import list_installed_skillslist_installed_skills(*, project_root: 'str | Path | None' = None) -> 'list[InstalledSkill]'load_agent_session
Stable · agent
from zhivex_ai import load_agent_sessionload_agent_session(agent: 'Agent', session_id: 'str', *, metadata: 'dict[str, Any] | None' = None) -> 'AgentSession'load_skill
Stable · skills
from zhivex_ai import load_skillload_skill(path: 'str | Path') -> 'SkillDefinition'load_skill_package
Beta · skills
from zhivex_ai import load_skill_packageload_skill_package(path: 'str | Path') -> 'SkillDefinition'mcp_http_server
Stable · agent
from zhivex_ai import mcp_http_servermcp_http_server(*, name: 'str', url: 'str', headers: 'dict[str, str] | None' = None, timeout_ms: 'int | None' = None) -> 'MCPServerConfig'mcp_stdio_server
Stable · agent
from zhivex_ai import mcp_stdio_servermcp_stdio_server(*, name: 'str', command: 'str', args: 'Iterable[str] | None' = None, env: 'dict[str, str] | None' = None, timeout_ms: 'int | None' = None) -> 'MCPServerConfig'meta_hosted_tool
Beta · provider
from zhivex_ai import meta_hosted_toolmeta_hosted_tool(tool_type: 'str', /, *, name: 'str | None' = None, tool_class: 'HostedToolClass | None' = None, **config: 'Any') -> 'HostedToolDefinition'meta_tool_search_tool
Beta · provider
from zhivex_ai import meta_tool_search_toolmeta_tool_search_tool(**config: 'Any') -> 'HostedToolDefinition'meta_web_search_tool
Beta · provider
from zhivex_ai import meta_web_search_toolmeta_web_search_tool(**config: 'Any') -> 'HostedToolDefinition'migrate_workflow_checkpoint
Stable · workflow
from zhivex_ai import migrate_workflow_checkpointmigrate_workflow_checkpoint(checkpoint: 'WorkflowCheckpoint', *, target_version: 'int' = 2, applied_at_ms: 'int | None' = None) -> 'WorkflowCheckpoint'Return a migrated checkpoint without mutating or persisting the source value.
Migrations are explicit, sequential, and auditable. Callers control the timestamp
when deterministic release or fixture evidence is required.migrate_workflow_checkpoint_payload
Stable · workflow
from zhivex_ai import migrate_workflow_checkpoint_payloadmigrate_workflow_checkpoint_payload(payload: 'dict[str, Any]', *, target_version: 'int' = 2, applied_at_ms: 'int | None' = None) -> 'dict[str, Any]'Deserialize, migrate, and serialize a persisted checkpoint payload.migrate_workflow_run_checkpoint
Stable · workflow
from zhivex_ai import migrate_workflow_run_checkpointmigrate_workflow_run_checkpoint(store: 'WorkflowCheckpointStore', run_id: 'str', *, target_version: 'int' = 2, applied_at_ms: 'int | None' = None) -> 'WorkflowCheckpoint'Append the migrated latest checkpoint using the store's compare-and-swap contract.open_websocket_connection
Experimental · realtime
from zhivex_ai import open_websocket_connectionopen_websocket_connection(url: 'str', *, headers: 'dict[str, str] | None' = None, options: 'RealtimeConnectOptions | None' = None) -> 'RealtimeConnection'openai_apply_patch_tool
Beta · provider
from zhivex_ai import openai_apply_patch_toolopenai_apply_patch_tool(**extra: 'Any') -> 'HostedToolDefinition'openai_code_interpreter_container
Beta · provider
from zhivex_ai import openai_code_interpreter_containeropenai_code_interpreter_container(*, file_ids: 'list[str] | None' = None, memory_limit: 'str | None' = None, network_policy: 'dict[str, Any] | None' = None, container_id: 'str | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_code_interpreter_tool
Beta · provider
from zhivex_ai import openai_code_interpreter_toolopenai_code_interpreter_tool(*, container: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_computer_use_tool
Beta · provider
from zhivex_ai import openai_computer_use_toolopenai_computer_use_tool(*, environment: 'str | None' = None, display_width: 'int | None' = None, display_height: 'int | None' = None, tool_type: 'str' = 'computer_use_preview', **extra: 'Any') -> 'HostedToolDefinition'openai_custom_tool
Beta · provider
from zhivex_ai import openai_custom_toolopenai_custom_tool(*, name: 'str', description: 'str | None' = None, format: 'dict[str, Any] | None' = None, defer_loading: 'bool | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_custom_tool_format_grammar
Beta · provider
from zhivex_ai import openai_custom_tool_format_grammaropenai_custom_tool_format_grammar(*, syntax: 'str', definition: 'str') -> 'dict[str, Any]'openai_custom_tool_format_text
Beta · provider
from zhivex_ai import openai_custom_tool_format_textopenai_custom_tool_format_text() -> 'dict[str, Any]'openai_domain_secret
Beta · provider
from zhivex_ai import openai_domain_secretopenai_domain_secret(*, domain: 'str', name: 'str', value: 'str') -> 'dict[str, Any]'openai_file_search_filter
Beta · provider
from zhivex_ai import openai_file_search_filteropenai_file_search_filter(*, key: 'str', operator: 'str', value: 'Any') -> 'dict[str, Any]'openai_file_search_filter_group
Beta · provider
from zhivex_ai import openai_file_search_filter_groupopenai_file_search_filter_group(operator: 'str', filters: 'list[dict[str, Any]]') -> 'dict[str, Any]'openai_file_search_tool
Beta · provider
from zhivex_ai import openai_file_search_toolopenai_file_search_tool(*, vector_store_ids: 'list[str]', filters: 'dict[str, Any] | None' = None, max_num_results: 'int | None' = None, ranking_options: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_hosted_tool
Beta · provider
from zhivex_ai import openai_hosted_toolopenai_hosted_tool(tool_type: 'str', /, *, name: 'str | None' = None, tool_class: 'str | None' = None, requires_approval: 'bool | None' = None, metadata: 'dict[str, Any] | None' = None, **config: 'Any') -> 'HostedToolDefinition'openai_image_generation_tool
Beta · provider
from zhivex_ai import openai_image_generation_toolopenai_image_generation_tool(*, model: 'str | None' = None, action: 'str | None' = None, background: 'str | None' = None, size: 'str | None' = None, quality: 'str | None' = None, output_format: 'str | None' = None, output_compression: 'int | None' = None, moderation: 'str | None' = None, partial_images: 'int | None' = None, input_fidelity: 'str | None' = None, input_image_mask: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_image_mask
Beta · provider
from zhivex_ai import openai_image_maskopenai_image_mask(*, file_id: 'str | None' = None, image_url: 'str | None' = None) -> 'dict[str, Any]'openai_inline_skill
Beta · provider
from zhivex_ai import openai_inline_skillopenai_inline_skill(*, name: 'str', description: 'str', source: 'dict[str, Any]') -> 'dict[str, Any]'openai_inline_skill_source
Beta · provider
from zhivex_ai import openai_inline_skill_sourceopenai_inline_skill_source(*, data: 'str', media_type: 'str' = 'application/zip') -> 'dict[str, Any]'openai_local_shell_tool
Experimental · provider
from zhivex_ai import openai_local_shell_toolopenai_local_shell_tool(**extra: 'Any') -> 'HostedToolDefinition'local shell execution is experimental and must be isolated by applications
openai_local_skill
Beta · provider
from zhivex_ai import openai_local_skillopenai_local_skill(*, name: 'str', path: 'str', description: 'str | None' = None) -> 'dict[str, Any]'openai_mcp_approval_response
Beta · provider
from zhivex_ai import openai_mcp_approval_responseopenai_mcp_approval_response(*, approval_request_id: 'str', approve: 'bool', id: 'str | None' = None, reason: 'str | None' = None) -> 'ProviderDataPart'openai_mcp_tool
Beta · provider
from zhivex_ai import openai_mcp_toolopenai_mcp_tool(*, server_url: 'str | None' = None, server_label: 'str | None' = None, headers: 'dict[str, str] | None' = None, allowed_tools: 'list[str] | None' = None, require_approval: 'str | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_namespace_tool
Beta · provider
from zhivex_ai import openai_namespace_toolopenai_namespace_tool(*, name: 'str', description: 'str', tools: 'list[dict[str, Any]]', **extra: 'Any') -> 'HostedToolDefinition'openai_network_policy_allowlist
Beta · provider
from zhivex_ai import openai_network_policy_allowlistopenai_network_policy_allowlist(*, allowed_domains: 'list[str]', domain_secrets: 'list[dict[str, Any]] | None' = None) -> 'dict[str, Any]'openai_network_policy_disabled
Beta · provider
from zhivex_ai import openai_network_policy_disabledopenai_network_policy_disabled() -> 'dict[str, Any]'openai_programmatic_tool_calling_tool
Beta · provider
from zhivex_ai import openai_programmatic_tool_calling_toolopenai_programmatic_tool_calling_tool(**extra: 'Any') -> 'HostedToolDefinition'openai_response_options
Beta · provider
from zhivex_ai import openai_response_optionsopenai_response_options(*, tools: 'list[HostedToolDefinition | dict[str, Any]] | None' = None, background: 'bool | None' = None, conversation: 'str | None' = None, previous_response_id: 'str | None' = None, previous_response: 'Any' = None, include: 'list[str] | None' = None, metadata: 'dict[str, Any] | None' = None, store: 'bool | None' = None, prompt: 'dict[str, Any] | None' = None, service_tier: 'str | None' = None, truncation: 'str | dict[str, Any] | None' = None, user: 'str | None' = None, safety_identifier: 'str | None' = None, reasoning: 'dict[str, Any] | None' = None, prompt_cache_options: 'dict[str, Any] | None' = None, multi_agent: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'dict[str, Any]'openai_response_reference
Beta · provider
from zhivex_ai import openai_response_referenceopenai_response_reference(*, response_id: 'str') -> 'ProviderDataPart'openai_shell_environment
Experimental · provider
from zhivex_ai import openai_shell_environmentopenai_shell_environment(*, file_ids: 'list[str] | None' = None, memory_limit: 'str | None' = None, network_policy: 'dict[str, Any] | None' = None, container_id: 'str | None' = None, local_skills: 'list[dict[str, Any]] | None' = None, use_local: 'bool' = False, **extra: 'Any') -> 'HostedToolDefinition'openai_shell_tool
Experimental · provider
from zhivex_ai import openai_shell_toolopenai_shell_tool(*, environment: 'HostedToolDefinition | dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'shell execution is experimental and must be isolated by applications
openai_skill_reference
Beta · provider
from zhivex_ai import openai_skill_referenceopenai_skill_reference(*, skill_id: 'str', version: 'str | None' = None) -> 'dict[str, Any]'openai_tool_search_tool
Beta · provider
from zhivex_ai import openai_tool_search_toolopenai_tool_search_tool(*, description: 'str | None' = None, execution: 'str | None' = None, parameters: 'dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'openai_user_location
Beta · provider
from zhivex_ai import openai_user_locationopenai_user_location(*, city: 'str | None' = None, country: 'str | None' = None, region: 'str | None' = None, timezone: 'str | None' = None) -> 'dict[str, Any]'openai_web_search_tool
Beta · provider
from zhivex_ai import openai_web_search_toolopenai_web_search_tool(*, search_context_size: 'str | None' = None, search_content_types: 'list[str] | None' = None, user_location: 'dict[str, Any] | None' = None, tool_type: 'str' = 'web_search', **extra: 'Any') -> 'HostedToolDefinition'parse_azure_openai_provider_data_part
Beta · provider
from zhivex_ai import parse_azure_openai_provider_data_partparse_azure_openai_provider_data_part(part: 'ProviderDataPart') -> 'AzureOpenAIProviderData | None'parse_openai_provider_data_part
Beta · provider
from zhivex_ai import parse_openai_provider_data_partparse_openai_provider_data_part(part: 'ProviderDataPart') -> 'OpenAIProviderData | None'parse_ui_message_request
Beta · transport
from zhivex_ai import parse_ui_message_requestparse_ui_message_request(request: 'Any', *, max_body_bytes: 'int' = 1048576, max_messages: 'int' = 1000) -> 'list[UIMessage]'permission_allowlist_approval_policy
Beta · agent
from zhivex_ai import permission_allowlist_approval_policypermission_allowlist_approval_policy(*allowed_permissions: 'str') -> 'ApprovalPolicy'prepare_subagents_for_agent
Beta · agent
from zhivex_ai import prepare_subagents_for_agentprepare_subagents_for_agent(agent: 'Agent') -> 'Agent'provider_data_part
Beta · messages
from zhivex_ai import provider_data_partprovider_data_part(provider: 'str', data: 'Any') -> 'ProviderDataPart'publish_skill
Beta · skills
from zhivex_ai import publish_skillpublish_skill(path: 'str | Path', *, registry_dir: 'str | Path') -> 'SkillRegistryIndex'qwen_code_interpreter_tool
Beta · provider
from zhivex_ai import qwen_code_interpreter_toolqwen_code_interpreter_tool(**config: 'Any') -> 'HostedToolDefinition'qwen_file_search_tool
Beta · provider
from zhivex_ai import qwen_file_search_toolqwen_file_search_tool(*, vector_store_ids: 'list[str]', **config: 'Any') -> 'HostedToolDefinition'qwen_hosted_tool
Beta · provider
from zhivex_ai import qwen_hosted_toolqwen_hosted_tool(type: 'str', *, name: 'str | None' = None, tool_class: 'HostedToolClass | None' = None, **config: 'Any') -> 'HostedToolDefinition'qwen_image_search_tool
Beta · provider
from zhivex_ai import qwen_image_search_toolqwen_image_search_tool(**config: 'Any') -> 'HostedToolDefinition'qwen_mcp_tool
Beta · provider
from zhivex_ai import qwen_mcp_toolqwen_mcp_tool(*, server_label: 'str', server_url: 'str', server_protocol: 'str' = 'sse', server_description: 'str | None' = None, headers: 'dict[str, str] | None' = None, **config: 'Any') -> 'HostedToolDefinition'qwen_web_extractor_tool
Beta · provider
from zhivex_ai import qwen_web_extractor_toolqwen_web_extractor_tool(**config: 'Any') -> 'HostedToolDefinition'qwen_web_search_image_tool
Beta · provider
from zhivex_ai import qwen_web_search_image_toolqwen_web_search_image_tool(**config: 'Any') -> 'HostedToolDefinition'qwen_web_search_tool
Beta · provider
from zhivex_ai import qwen_web_search_toolqwen_web_search_tool(**config: 'Any') -> 'HostedToolDefinition'remote_tool
Beta · messages
from zhivex_ai import remote_toolremote_tool(*, name: 'str', url: 'str', schema: 'Any', description: 'str | None' = None, headers: 'dict[str, str] | None' = None, timeout_ms: 'int | None' = None, tags: 'list[str] | None' = None, requires_approval: 'bool | None' = None, permissions: 'list[str] | None' = None, metadata: 'dict[str, Any] | None' = None, input_guardrails: 'list[ToolInputGuardrail] | None' = None, output_guardrails: 'list[ToolOutputGuardrail] | None' = None) -> 'ToolDefinition'render_provider_support_markdown
Beta · provider-support
from zhivex_ai import render_provider_support_markdownrender_provider_support_markdown(rows: 'Iterable[ProviderSupportRow]') -> 'str'replay_agent_run
Stable · agent
from zhivex_ai import replay_agent_runreplay_agent_run(state: 'AgentRunState') -> 'AgentReplayResult'resume_agent
Stable · agent
from zhivex_ai import resume_agentresume_agent(*, agent: 'Agent[AgentDepsT, AgentOutputT]', session_id: 'str', prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, deps: 'AgentDepsT | None' = None, tools: 'ToolSet | ToolRegistry | None' = None, skills: 'SkillSet | SkillRegistry | None' = None, tool_choice: 'str | ToolChoiceName | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, max_steps: 'int | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, stop_on_handoff: 'bool' = False, runtime: 'AgentRuntime | None' = None, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None) -> 'Awaitable[AgentRunResult[AgentOutputT]]'resume_agent_run
Stable · agent
from zhivex_ai import resume_agent_runresume_agent_run(*, agent: 'Agent[AgentDepsT, AgentOutputT]', run_id: 'str', approval_id: 'str | None' = None, approved: 'bool' = True, reason: 'str | None' = None, deps: 'AgentDepsT | None' = None, tools: 'ToolSet | ToolRegistry | None' = None, skills: 'SkillSet | SkillRegistry | None' = None, tool_choice: 'str | ToolChoiceName | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, max_steps: 'int | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, stop_on_handoff: 'bool' = False, runtime: 'AgentRuntime | None' = None, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, idempotency_key: 'str | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None) -> 'Awaitable[AgentRunResult[AgentOutputT]]'resume_workflow
Stable · workflow
from zhivex_ai import resume_workflowresume_workflow(workflow: 'WorkflowGraph', run_id: 'str', *, interrupt_id: 'str | None' = None, resume_value: 'JsonValue | None' = None, state_updates: 'Mapping[str, JsonValue] | None' = None, approval_id: 'str | None' = None, approved: 'bool' = True, reason: 'str | None' = None, node_name: 'str | None' = None, deps: 'Any' = None, session: 'AgentSession | None' = None) -> 'WorkflowRunResult'run_agent
Stable · agent
from zhivex_ai import run_agentrun_agent(*, agent: 'Agent[AgentDepsT, AgentOutputT]', session: 'AgentSession | None' = None, prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, deps: 'AgentDepsT | None' = None, tools: 'ToolSet | ToolRegistry | None' = None, skills: 'SkillSet | SkillRegistry | None' = None, tool_choice: 'str | ToolChoiceName | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, max_steps: 'int | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, stop_on_handoff: 'bool' = False, runtime: 'AgentRuntime | None' = None, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None) -> 'Awaitable[AgentRunResult[AgentOutputT]]'run_agent_evaluation
Beta · agent
from zhivex_ai import run_agent_evaluationrun_agent_evaluation(*, agent: 'Agent | AgentEvaluationAgentFactory', dataset: 'list[AgentEvaluationCase]', repetitions: 'int' = 1, max_concurrency: 'int' = 1, cost_estimator: 'AgentEvaluationCostEstimator | None' = None) -> 'AgentEvaluationResult'Run a dataset with bounded concurrency while preserving input order.run_agent_evaluation_experiment
Beta · agent
from zhivex_ai import run_agent_evaluation_experimentrun_agent_evaluation_experiment(*, variants: 'list[AgentEvaluationVariant] | dict[str, Agent | AgentEvaluationAgentFactory]', dataset: 'list[AgentEvaluationCase]', baseline: 'str | None' = None, metrics: 'list[AgentEvaluationMetric] | None' = None, gates: 'list[AgentEvaluationGate] | None' = None, metadata: 'dict[str, JsonValue] | None' = None, repetitions: 'int' = 1, max_concurrency: 'int' = 1, cost_estimator: 'AgentEvaluationCostEstimator | None' = None) -> 'AgentEvaluationExperimentResult'Evaluate agent variants deterministically and apply baseline-aware CI gates.
Variants, cases, and custom metrics run in their supplied order. Custom metric
values are averaged across cases, and every emitted number is required to be
finite so ``to_json`` always produces strict JSON.run_agent_evaluation_fixture
Beta · agent
from zhivex_ai import run_agent_evaluation_fixturerun_agent_evaluation_fixture(fixture: 'AgentEvaluationFixture', *, agent: 'Agent | Callable[[AgentEvaluationCase], Agent | Awaitable[Agent]]', repetitions: 'int' = 1, max_concurrency: 'int' = 1, cost_estimator: 'AgentEvaluationCostEstimator | None' = None) -> 'AgentEvaluationResult'run_agent_group
Beta · agent
from zhivex_ai import run_agent_grouprun_agent_group(members: 'list[AgentGroupMember]', *, prompt: 'str | None' = None, parent_run_id: 'str | None' = None, deps: 'Any' = None, runtime: 'AgentRuntime | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None, max_concurrency: 'int | None' = None, timeout_ms: 'int | None' = None, fail_fast: 'bool' = False, idempotency_key: 'str | None' = None, cancellation_token: 'AgentCancellationToken | None' = None) -> 'AgentGroupRunResult'run_skill
Beta · skills
from zhivex_ai import run_skillrun_skill(name: 'str', *, entrypoint: 'str | None' = None, input: 'Any' = None, project_root: 'str | Path | None' = None) -> 'SkillRunResult'run_workflow
Stable · workflow
from zhivex_ai import run_workflowrun_workflow(workflow: 'WorkflowAgent', *, session: 'AgentSession | None' = None, prompt: 'str | None' = None, parent_run_id: 'str | None' = None) -> 'WorkflowRunResult'serialize_agent_run_state
Stable · agent
from zhivex_ai import serialize_agent_run_stateserialize_agent_run_state(state: 'AgentRunState') -> 'dict[str, Any]'serialize_ui_message
Beta · ui
from zhivex_ai import serialize_ui_messageserialize_ui_message(message: 'UIMessage') -> 'str'serialize_ui_message_chunk
Beta · ui
from zhivex_ai import serialize_ui_message_chunkserialize_ui_message_chunk(chunk: 'UIMessageChunk') -> 'str'serialize_workflow_checkpoint
Stable · workflow
from zhivex_ai import serialize_workflow_checkpointserialize_workflow_checkpoint(checkpoint: 'WorkflowCheckpoint') -> 'dict[str, Any]'set_agent_session_skills
Stable · agent
from zhivex_ai import set_agent_session_skillsset_agent_session_skills(session: 'AgentSession', *skill_names: 'str') -> 'AgentSession'skill
Stable · skills
from zhivex_ai import skillskill(definition: 'SkillDefinition | None' = None, *, name: 'str | None' = None, description: 'str | None' = None, instructions: 'str | None' = None, path: 'str | None' = None, display_name: 'str | None' = None, short_description: 'str | None' = None, default_prompt: 'str | None' = None, allow_implicit_invocation: 'bool | None' = None, priority: 'int | None' = None, triggers: 'list[str] | None' = None, anti_triggers: 'list[str] | None' = None, allowed_providers: 'list[str] | None' = None, allowed_models: 'list[str] | None' = None, persist_to_session: 'bool | None' = None, dependency_failure_mode: 'SkillDependencyFailureMode | None' = None, tools: 'dict[str, ToolDefinition] | None' = None, dependencies: 'list[SkillDependency] | None' = None, metadata: 'dict[str, Any] | None' = None, version: 'str | None' = None, entrypoints: 'list[SkillEntrypoint] | None' = None, artifacts: 'list[SkillArtifact] | None' = None, permissions: 'SkillPermissions | None' = None, resources: 'list[str] | None' = None, source: 'str | None' = None, checksum: 'str | None' = None, package_manifest: 'SkillPackageManifest | None' = None, package_manifest_path: 'str | None' = None, install_path: 'str | None' = None) -> 'SkillDefinition'stream_agent
Stable · agent
from zhivex_ai import stream_agentstream_agent(*, agent: 'Agent[AgentDepsT, AgentOutputT]', session: 'AgentSession | None' = None, prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, deps: 'AgentDepsT | None' = None, tools: 'ToolSet | ToolRegistry | None' = None, skills: 'SkillSet | SkillRegistry | None' = None, tool_choice: 'str | ToolChoiceName | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, max_steps: 'int | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, stop_on_handoff: 'bool' = False, runtime: 'AgentRuntime | None' = None, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, idempotency_key: 'str | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None, stream_buffer_size: 'int | None' = None) -> 'AgentStreamResult[AgentOutputT]'stream_agent_ag_ui
Beta · protocol
from zhivex_ai import stream_agent_ag_uistream_agent_ag_ui(*, agent: 'Agent', prompt: 'str', thread_id: 'str', run_id: 'str | None' = None, run_options: 'HostedAgentRunOptions | None' = None, run_options_resolver: 'ProtocolRunOptionsResolver | None' = None, error_mapper: 'ProtocolErrorMapper | None' = None, on_protocol_event: 'ProtocolEventCallback | None' = None, limits: 'ProtocolLimits | None' = None) -> 'AsyncIterable[dict[str, Any]]'Translate a Zhivex agent stream into canonical AG-UI lifecycle events.stream_live_agent
Experimental · agent
from zhivex_ai import stream_live_agentstream_live_agent(*, agent: 'Agent[AgentDepsT, AgentOutputT]', session: 'AgentSession | None' = None, deps: 'AgentDepsT | None' = None, tools: 'ToolSet | ToolRegistry | None' = None, skills: 'SkillSet | SkillRegistry | None' = None, tool_choice: 'str | ToolChoiceName | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, connect_options: 'RealtimeConnectOptions | None' = None, realtime_config: 'RealtimeSessionConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, runtime: 'AgentRuntime | None' = None, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None, stream_buffer_size: 'int | None' = None) -> 'LiveAgentStreamResult[AgentOutputT]'realtime/live APIs are experimental
stream_object
Stable · foundation
from zhivex_ai import stream_objectstream_object(*, model: 'Any', schema: 'Any', prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, system: 'str | None' = None, mode: 'str' = 'auto', schema_name: 'str | None' = None, schema_description: 'str | None' = None, stream_buffer_size: 'int | None' = None, **kwargs: 'Any') -> '_StreamObjectResult'stream_sse
Stable · transport
from zhivex_ai import stream_ssestream_sse(response: 'ResponseLike')stream_text
Stable · foundation
from zhivex_ai import stream_textstream_text(*, model: 'LanguageModel', prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, system: 'str | None' = None, tools: 'ToolSet | None' = None, tool_choice: 'ToolChoice | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, max_steps: 'int | None' = None, temperature: 'float | None' = None, max_tokens: 'int | None' = None, reasoning: 'ReasoningConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, retrieval: 'PortableRetrievalConfig | None' = None, structured_output: 'Any' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, stream_buffer_size: 'int | None' = None, on_event: 'Callable[[StreamEvent], Awaitable[None]] | None' = None) -> '_StreamTextResult'summarize_agent_trace
Beta · observability
from zhivex_ai import summarize_agent_tracesummarize_agent_trace(state_or_trace: 'AgentRunState | AgentTraceArtifact', *, pricing: 'TokenPricing | ModelCatalog | None' = None) -> 'AgentTraceSummary'system
Beta · messages
from zhivex_ai import systemsystem(text: 'str') -> 'ModelMessage'text_part
Beta · messages
from zhivex_ai import text_parttext_part(text: 'str') -> 'TextPart'to_ag_ui_sse_response
Beta · protocol
from zhivex_ai import to_ag_ui_sse_responseto_ag_ui_sse_response(source: 'AsyncIterable[dict[str, Any]]') -> 'HTTPResponse'Encode AG-UI events with the official protocol encoder.to_sse_response
Stable · transport
from zhivex_ai import to_sse_responseto_sse_response(source: 'AsyncIterable[Any]', *, event: 'str | Callable[[Any], str | None] | None' = None, status_code: 'int' = 200, headers: 'Mapping[str, str] | None' = None) -> 'HTTPResponse'to_sse_stream
Stable · transport
from zhivex_ai import to_sse_streamto_sse_stream(source: 'AsyncIterable[Any]', *, event: 'str | Callable[[Any], str | None] | None' = None)to_text_stream
Stable · transport
from zhivex_ai import to_text_streamto_text_stream(result: 'Any')to_text_stream_response
Stable · transport
from zhivex_ai import to_text_stream_responseto_text_stream_response(result: 'Any', *, status_code: 'int' = 200, headers: 'Mapping[str, str] | None' = None) -> 'HTTPResponse'to_ui_message
Beta · ui
from zhivex_ai import to_ui_messageto_ui_message(message: 'ModelMessage', id: 'str | None' = None) -> 'UIMessage'to_ui_message_stream
Beta · ui
from zhivex_ai import to_ui_message_streamto_ui_message_stream(source: 'Any', message_id: 'str | None' = None)to_ui_message_stream_response
Stable · transport
from zhivex_ai import to_ui_message_stream_responseto_ui_message_stream_response(source: 'Any', *, message_id: 'str | None' = None, status_code: 'int' = 200, headers: 'Mapping[str, str] | None' = None) -> 'HTTPResponse'to_ui_messages
Beta · ui
from zhivex_ai import to_ui_messagesto_ui_messages(messages: 'list[ModelMessage]') -> 'list[UIMessage]'tool
Stable · messages
from zhivex_ai import tooltool(definition: 'ToolDefinition | Callable[..., Any] | None' = None, *, name: 'str | None' = None, description: 'str | None' = None, schema: 'Any' = None, execute: 'Callable[..., Any] | None' = None, input_examples: 'list[Any] | None' = None, strict: 'bool | None' = None, defer_loading: 'bool | None' = None, eager_input_streaming: 'bool | None' = None, allowed_callers: 'list[str] | None' = None, output_schema: 'Any' = None, cache_control: 'dict[str, Any] | None' = None, tags: 'list[str] | None' = None, requires_approval: 'bool | None' = None, permissions: 'list[str] | None' = None, source: 'ToolSource' = 'local', metadata: 'dict[str, Any] | None' = None, supports_streaming: 'bool' = False, remote_config: 'RemoteHTTPToolConfig | None' = None, mcp_config: 'MCPToolConfig | None' = None, input_guardrails: 'list[ToolInputGuardrail] | None' = None, output_guardrails: 'list[ToolOutputGuardrail] | None' = None, **kwargs: 'Any') -> 'ToolDefinition | ToolDecorator'tool_call_part
Beta · messages
from zhivex_ai import tool_call_parttool_call_part(tool_call: 'ToolCall') -> 'ToolCallPart'tool_result_part
Beta · messages
from zhivex_ai import tool_result_parttool_result_part(tool_result: 'ToolExecutionResult') -> 'ToolResultPart'transcribe_audio
Beta · foundation
from zhivex_ai import transcribe_audiotranscribe_audio(*, model: 'TranscriptionModel', audio: 'AudioInput', prompt: 'str | None' = None, language: 'str | None' = None, config: 'PortableTranscriptionConfig | None' = None, provider_options: 'dict[str, object] | None' = None, timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None) -> 'TranscriptionOutput'user
Beta · messages
from zhivex_ai import useruser(input: 'str | list[ContentPart]') -> 'ModelMessage'validate_skill
Beta · skills
from zhivex_ai import validate_skillvalidate_skill(path: 'str | Path') -> 'SkillDefinition'validate_workflow_expectations
Stable · workflow
from zhivex_ai import validate_workflow_expectationsvalidate_workflow_expectations(result: 'WorkflowRunResult', expectations: 'object') -> 'list[str]'vertex_code_execution_tool
Beta · provider
from zhivex_ai import vertex_code_execution_toolvertex_code_execution_tool(**config: 'Any') -> 'HostedToolDefinition'vertex_computer_use_tool
Beta · provider
from zhivex_ai import vertex_computer_use_toolvertex_computer_use_tool(**config: 'Any') -> 'HostedToolDefinition'vertex_external_search_tool
Beta · provider
from zhivex_ai import vertex_external_search_toolvertex_external_search_tool(*, endpoint: 'str', api_key: 'str', api_spec: 'str' = 'SIMPLE_SEARCH', **extra: 'Any') -> 'dict[str, Any]'vertex_google_maps_tool
Beta · provider
from zhivex_ai import vertex_google_maps_toolvertex_google_maps_tool(**config: 'Any') -> 'HostedToolDefinition'vertex_google_search_tool
Beta · provider
from zhivex_ai import vertex_google_search_toolvertex_google_search_tool(*, exclude_domains: 'list[str] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'vertex_url_context_tool
Beta · provider
from zhivex_ai import vertex_url_context_toolvertex_url_context_tool(**config: 'Any') -> 'HostedToolDefinition'vertex_vertex_ai_search_tool
Beta · provider
from zhivex_ai import vertex_vertex_ai_search_toolvertex_vertex_ai_search_tool(*, datastore: 'str', **extra: 'Any') -> 'dict[str, Any]'workflow_checkpoint_from_json
Stable · workflow
from zhivex_ai import workflow_checkpoint_from_jsonworkflow_checkpoint_from_json(value: 'str') -> 'WorkflowCheckpoint'workflow_checkpoint_to_json
Stable · workflow
from zhivex_ai import workflow_checkpoint_to_jsonworkflow_checkpoint_to_json(checkpoint: 'WorkflowCheckpoint') -> 'str'workflow_step
Stable · workflow
from zhivex_ai import workflow_stepworkflow_step(name: 'str', agent: 'Agent | None' = None, *, prompt: 'str | None' = None, input_template: 'str | None' = None, output_key: 'str | None' = None, metadata_key: 'str | None' = None, max_retries: 'int | None' = None, timeout_ms: 'int | None' = None, error_policy: 'WorkflowErrorPolicy' = 'fail_fast', retry_policy: 'WorkflowRetryPolicy | None' = None, idempotency_key: 'str | None' = None, executor_ref: 'str | None' = None, metadata: 'dict[str, JsonValue] | None' = None, executor: 'WorkflowFunctionExecutor | None' = None) -> 'WorkflowStep'wrap_language_model
Beta · middleware
from zhivex_ai import wrap_language_modelwrap_language_model(model: 'LanguageModel', middlewares: 'list[GenerateMiddleware]') -> 'LanguageModel'