> ## Documentation Index > Fetch the complete documentation index at: https://arizeai-433a7140.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Chat ## OpenAPI ````yaml post /v1/agent_sessions/{session_id}/chat openapi: 3.1.0 info: title: Arize-Phoenix REST API description: Schema for Arize-Phoenix REST API version: '1.0' servers: [] security: [] paths: /v1/agent_sessions/{session_id}/chat: post: tags: - chat summary: Chat operationId: agentSessionChat parameters: - name: session_id in: path required: true schema: type: string title: Session Id requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ChatRequestBody' responses: '200': description: Successful Response content: application/json: schema: {} '400': content: text/plain: schema: type: string description: Bad Request '401': content: text/plain: schema: type: string description: Unauthorized '403': content: text/plain: schema: type: string description: Forbidden '404': content: text/plain: schema: type: string description: Not Found '409': description: >- The request conflicts with the session's current state; the body's ``code`` field says how. content: application/json: schema: $ref: '#/components/schemas/AgentSessionConflictError' '422': description: Validation Error content: application/json: schema: $ref: '#/components/schemas/HTTPValidationError' '507': content: text/plain: schema: type: string description: Insufficient Storage components: schemas: ChatRequestBody: properties: headless: type: boolean title: Headless description: >- Whether a headless client (terminal or scripted) is driving the turn, as opposed to the browser assistant. Selects the agent configuration the turn runs on. contexts: items: $ref: '#/components/schemas/ChatContext' type: array title: Contexts editPermission: type: string enum: - manual - bypass title: Editpermission default: manual requestedSkills: items: type: string type: array title: Requestedskills description: >- Skills the user explicitly requested via the prompt's slash-command affordance. The server force-loads each available skill by injecting a synthetic load_skill tool call/result at the tail of the message history. Unknown or context-unavailable names are ignored. model: $ref: '#/components/schemas/AgentModelSelection' description: >- The model the client believes the session is set to. This is a precondition, not an instruction: the turn always runs on the session's persisted selection, and a mismatch is rejected with HTTP 409 and code ``agent_session_model_stale`` rather than silently running on — or switching to — an unexpected model. Change the session's model with ``PATCH .../agent_sessions/{session_id}``. trigger: type: string const: submit-message title: Trigger default: submit-message id: type: string title: Id message: anyOf: - $ref: '#/components/schemas/PhoenixUIMessage' - type: 'null' description: >- The turn's new user message to append. May be omitted for client-tool continuation, where ``toolOutputs`` resolve the trailing assistant message's pending tool calls instead. toolOutputs: items: anyOf: - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__ToolOutputAvailablePart - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__ToolOutputErrorPart - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__DynamicToolOutputAvailablePart - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__DynamicToolOutputErrorPart type: array title: Tooloutputs description: >- Client-executed tool results for pending tool calls on the transcript's trailing assistant message, matched by ``toolCallId``. Submitted alone they continue the assistant turn; submitted with ``message`` they resolve dangling tool calls before the new user turn runs. lastMessageId: anyOf: - type: string - type: 'null' title: Lastmessageid description: >- The id of the last transcript message the client has rendered, used for optimistic concurrency. Omit when the session has no messages; required (and validated against the persisted transcript) once it does. On mismatch the server rejects the send with HTTP 409 and code ``agent_session_messages_stale`` — the client should refetch the session before retrying. recordLocalTraces: type: boolean title: Recordlocaltraces default: false exportRemoteTraces: type: boolean title: Exportremotetraces default: false instrumentUserId: type: boolean title: Instrumentuserid description: >- When true and the request is authenticated as a PhoenixUser, attaches the user's email as the OpenInference ``user.id`` span attribute on all traced work for this request. default: false type: object required: - headless - model - id title: ChatRequestBody description: Assistant chat submit request payload. AgentSessionConflictError: properties: code: type: string enum: - agent_session_busy - agent_session_model_stale - agent_session_messages_stale - agent_session_tool_outputs_conflict - agent_session_already_compact - agent_session_compaction_conflict title: Code description: Machine-readable reason the request conflicted. message: anyOf: - type: string - type: 'null' title: Message description: Optional human-readable elaboration on the conflict. type: object required: - code title: AgentSessionConflictError description: >- Body of every HTTP 409 returned by the agent session routes. - ``agent_session_busy``: another turn holds the session's turn lock. - ``agent_session_model_stale``: the request asserted a model the session is no longer set to; refetch the session before retrying. - ``agent_session_messages_stale``: the send's ``lastMessageId`` no longer matches the persisted transcript — another client appended; refetch the transcript and retry. - ``agent_session_tool_outputs_conflict``: the submitted ``toolOutputs`` do not match the transcript's trailing assistant message (no trailing assistant message to continue, an unknown ``toolCallId``, or a tool-name mismatch). Unlike ``agent_session_messages_stale`` this is not a concurrent-writer race but an inconsistent request; fix the client rather than retrying. - ``agent_session_already_compact``: there are no complete turns to compact — either nothing new has finished since the transcript's latest checkpoint, or a concurrent request's checkpoint already covers them. Not retryable; the conversation is as compact as it can get. - ``agent_session_compaction_conflict``: the conversation changed while it was being compacted; retry. HTTPValidationError: properties: detail: items: $ref: '#/components/schemas/ValidationError' type: array title: Detail type: object title: HTTPValidationError ChatContext: oneOf: - $ref: '#/components/schemas/AppContext' - $ref: '#/components/schemas/ProjectContext' - $ref: '#/components/schemas/TraceContext' - $ref: '#/components/schemas/SessionContext' - $ref: '#/components/schemas/PromptContext' - $ref: '#/components/schemas/PromptVersionContext' - $ref: '#/components/schemas/AgentSpanContext' - $ref: '#/components/schemas/PlaygroundContext' - $ref: '#/components/schemas/CodeEvaluatorContext' - $ref: '#/components/schemas/LlmEvaluatorContext' - $ref: '#/components/schemas/DatasetContext' - $ref: '#/components/schemas/GraphQLContext' - $ref: '#/components/schemas/WebAccessContext' - $ref: '#/components/schemas/SubagentsContext' title: ChatContext description: Discriminated union of every UI-state context the agent understands. discriminator: propertyName: type mapping: app: $ref: '#/components/schemas/AppContext' code_evaluator: $ref: '#/components/schemas/CodeEvaluatorContext' dataset: $ref: '#/components/schemas/DatasetContext' graphql: $ref: '#/components/schemas/GraphQLContext' llm_evaluator: $ref: '#/components/schemas/LlmEvaluatorContext' playground: $ref: '#/components/schemas/PlaygroundContext' project: $ref: '#/components/schemas/ProjectContext' prompt: $ref: '#/components/schemas/PromptContext' prompt_version: $ref: '#/components/schemas/PromptVersionContext' session: $ref: '#/components/schemas/SessionContext' span: $ref: '#/components/schemas/AgentSpanContext' subagents: $ref: '#/components/schemas/SubagentsContext' trace: $ref: '#/components/schemas/TraceContext' web_access: $ref: '#/components/schemas/WebAccessContext' AgentModelSelection: oneOf: - $ref: '#/components/schemas/CustomProviderModelSelection' - $ref: '#/components/schemas/BuiltInProviderModelSelection' discriminator: propertyName: providerType mapping: builtin: $ref: '#/components/schemas/BuiltInProviderModelSelection' custom: $ref: '#/components/schemas/CustomProviderModelSelection' PhoenixUIMessage: properties: id: type: string title: Id role: type: string enum: - system - user - assistant title: Role metadata: anyOf: - $ref: '#/components/schemas/MessageMetadata' - type: 'null' parts: items: anyOf: - $ref: '#/components/schemas/TextUIPart' - $ref: '#/components/schemas/ReasoningUIPart' - $ref: '#/components/schemas/ToolInputStreamingPart' - $ref: '#/components/schemas/ToolInputAvailablePart' - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__ToolOutputAvailablePart - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__ToolOutputErrorPart - $ref: '#/components/schemas/ToolApprovalRequestedPart' - $ref: '#/components/schemas/ToolApprovalRespondedPart' - $ref: '#/components/schemas/ToolOutputDeniedPart' - $ref: '#/components/schemas/DynamicToolInputStreamingPart' - $ref: '#/components/schemas/DynamicToolInputAvailablePart' - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__DynamicToolOutputAvailablePart - $ref: >- #/components/schemas/phoenix__db__types__data_stream_protocol__request_types__DynamicToolOutputErrorPart - $ref: '#/components/schemas/DynamicToolApprovalRequestedPart' - $ref: '#/components/schemas/DynamicToolApprovalRespondedPart' - $ref: '#/components/schemas/DynamicToolOutputDeniedPart' - $ref: '#/components/schemas/SourceUrlUIPart' - $ref: '#/components/schemas/SourceDocumentUIPart' - $ref: '#/components/schemas/FileUIPart' - $ref: '#/components/schemas/DataUIPart' - $ref: '#/components/schemas/StepStartUIPart' type: array title: Parts additionalProperties: false type: object required: - id - role - parts title: PhoenixUIMessage description: '``UIMessage`` with metadata narrowed to the Phoenix wire shapes.' phoenix__db__types__data_stream_protocol__request_types__ToolOutputAvailablePart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: output-available title: State default: output-available input: anyOf: - {} - type: 'null' title: Input output: anyOf: - {} - type: 'null' title: Output providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata resultProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Resultprovidermetadata preliminary: anyOf: - type: boolean - type: 'null' title: Preliminary approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId title: ToolOutputAvailablePart description: Tool part in output-available state. phoenix__db__types__data_stream_protocol__request_types__ToolOutputErrorPart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: output-error title: State default: output-error input: anyOf: - {} - type: 'null' title: Input rawInput: anyOf: - {} - type: 'null' title: Rawinput errorText: type: string title: Errortext providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata resultProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Resultprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId - errorText title: ToolOutputErrorPart description: Tool part in output-error state. phoenix__db__types__data_stream_protocol__request_types__DynamicToolOutputAvailablePart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: output-available title: State default: output-available input: title: Input output: title: Output providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata resultProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Resultprovidermetadata preliminary: anyOf: - type: boolean - type: 'null' title: Preliminary approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId - input - output title: DynamicToolOutputAvailablePart description: Dynamic tool part in output-available state. phoenix__db__types__data_stream_protocol__request_types__DynamicToolOutputErrorPart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: output-error title: State default: output-error input: title: Input errorText: type: string title: Errortext providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata resultProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Resultprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId - input - errorText title: DynamicToolOutputErrorPart description: Dynamic tool part in output-error state. ValidationError: properties: loc: items: anyOf: - type: string - type: integer type: array title: Location msg: type: string title: Message type: type: string title: Error Type input: title: Input ctx: type: object title: Context type: object required: - loc - msg - type title: ValidationError AppContext: properties: type: type: string const: app title: Type currentDateTime: type: string title: Currentdatetime timeZone: type: string title: Timezone type: object required: - type - currentDateTime - timeZone title: AppContext description: Per-turn browser clock context for resolving relative time requests. ProjectContext: properties: type: type: string const: project title: Type projectNodeId: type: string title: Projectnodeid spanFilter: anyOf: - type: string - type: 'null' title: Spanfilter type: object required: - type - projectNodeId title: ProjectContext description: >- Project the user is currently viewing. ``span_filter`` carries the project-scoped span filter expression when the span filter field is mounted — empty string when the field is mounted with no condition applied, ``None`` when the field is not present at all. It describes the view in full, root-span scoping included (which is expressed within the filter DSL as ``parent_id is None``). TraceContext: properties: type: type: string const: trace title: Type projectNodeId: type: string title: Projectnodeid otelTraceId: type: string title: Oteltraceid type: object required: - type - projectNodeId - otelTraceId title: TraceContext SessionContext: properties: type: type: string const: session title: Type projectNodeId: type: string title: Projectnodeid sessionNodeId: type: string title: Sessionnodeid type: object required: - type - projectNodeId - sessionNodeId title: SessionContext description: Session the user is currently viewing. PromptContext: properties: type: type: string const: prompt title: Type promptNodeId: type: string title: Promptnodeid type: object required: - type - promptNodeId title: PromptContext description: Prompt the user is currently viewing. PromptVersionContext: properties: type: type: string const: prompt_version title: Type promptNodeId: type: string title: Promptnodeid promptVersionNodeId: type: string title: Promptversionnodeid type: object required: - type - promptNodeId - promptVersionNodeId title: PromptVersionContext description: Prompt version the user is currently viewing. AgentSpanContext: properties: type: type: string const: span title: Type projectNodeId: anyOf: - type: string - type: 'null' title: Projectnodeid spanNodeId: anyOf: - type: string - type: 'null' title: Spannodeid otelSpanId: anyOf: - type: string - type: 'null' title: Otelspanid type: object required: - type title: AgentSpanContext description: >- Span the user has selected. Exactly one of ``span_node_id`` (relay) or ``otel_span_id`` (OpenTelemetry hex) must be set. ``project_node_id`` is optional because a span can be selected from views outside a project route. PlaygroundContext: properties: type: type: string const: playground title: Type recordExperiments: type: boolean title: Recordexperiments default: true repetitions: type: integer title: Repetitions default: 1 nextExperimentScaffold: anyOf: - $ref: '#/components/schemas/PlaygroundExperimentScaffoldContext' - type: 'null' instances: items: $ref: '#/components/schemas/PlaygroundInstanceContext' type: array title: Instances evaluators: items: $ref: '#/components/schemas/PlaygroundEvaluatorContext' type: array title: Evaluators type: object required: - type title: PlaygroundContext description: Playground prompt editor state mounted in the current browser route. CodeEvaluatorContext: properties: type: type: string const: code_evaluator title: Type evaluatorNodeId: anyOf: - type: string - type: 'null' title: Evaluatornodeid type: object required: - type title: CodeEvaluatorContext description: Code-evaluator create/edit form mounted in the current browser route. LlmEvaluatorContext: properties: type: type: string const: llm_evaluator title: Type evaluatorNodeId: anyOf: - type: string - type: 'null' title: Evaluatornodeid type: object required: - type title: LlmEvaluatorContext description: LLM-evaluator create/edit form mounted in the current browser route. DatasetContext: properties: type: type: string const: dataset title: Type datasetNodeId: type: string title: Datasetnodeid datasetVersionNodeId: anyOf: - type: string - type: 'null' title: Datasetversionnodeid type: object required: - type - datasetNodeId title: DatasetContext description: >- Dataset the user is currently viewing or has bound to a workflow. Carries the dataset's relay node id and, when known, the active version node id. These IDs scope the create-form handoff link and the sampling of active dataset examples used as prompt context; the dataset schema itself is open. GraphQLContext: properties: type: type: string const: graphql title: Type mutationsEnabled: type: boolean title: Mutationsenabled type: object required: - type - mutationsEnabled title: GraphQLContext description: GraphQL runtime state. WebAccessContext: properties: type: type: string const: web_access title: Type enabled: type: boolean title: Enabled type: object required: - type - enabled title: WebAccessContext description: User's per-turn request to expose web search / fetch tools. SubagentsContext: properties: type: type: string const: subagents title: Type enabled: type: boolean title: Enabled type: object required: - type - enabled title: SubagentsContext description: User's per-turn request to expose the subagent-spawning tool. CustomProviderModelSelection: properties: providerType: type: string const: custom title: Providertype providerId: type: string title: Providerid modelName: type: string title: Modelname type: object required: - providerType - providerId - modelName title: CustomProviderModelSelection description: Chat against a stored custom provider record. BuiltInProviderModelSelection: properties: providerType: type: string const: builtin title: Providertype provider: $ref: '#/components/schemas/ModelProvider' modelName: type: string title: Modelname type: object required: - providerType - provider - modelName title: BuiltInProviderModelSelection description: |- Chat against a Phoenix built-in provider. Credentials and connection details (base URL, Azure endpoint, AWS region) are resolved from the secret store first and the process environment second. MessageMetadata: properties: phoenix: anyOf: - oneOf: - $ref: '#/components/schemas/PhoenixAssistantMessageMetadata' - $ref: '#/components/schemas/PhoenixUserMessageMetadata' discriminator: propertyName: type mapping: assistant: $ref: '#/components/schemas/PhoenixAssistantMessageMetadata' user: $ref: '#/components/schemas/PhoenixUserMessageMetadata' - type: 'null' title: Phoenix pydantic_ai: anyOf: - $ref: '#/components/schemas/PydanticAIMessageMetadata' - type: 'null' additionalProperties: false type: object title: MessageMetadata description: '``UIMessage.metadata`` as a registry of namespaces.' TextUIPart: properties: type: type: string const: text title: Type default: text text: type: string title: Text state: anyOf: - type: string enum: - streaming - done - type: 'null' title: State providerMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Providermetadata additionalProperties: false type: object required: - text title: TextUIPart description: A text part of a message. ReasoningUIPart: properties: type: type: string const: reasoning title: Type default: reasoning text: type: string title: Text state: anyOf: - type: string enum: - streaming - done - type: 'null' title: State providerMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Providermetadata additionalProperties: false type: object required: - text title: ReasoningUIPart description: A reasoning part of a message. ToolInputStreamingPart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: input-streaming title: State default: input-streaming input: anyOf: - {} - type: 'null' title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId title: ToolInputStreamingPart description: Tool part in input-streaming state. ToolInputAvailablePart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: input-available title: State default: input-available input: anyOf: - {} - type: 'null' title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId title: ToolInputAvailablePart description: Tool part in input-available state. ToolApprovalRequestedPart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: approval-requested title: State default: approval-requested input: anyOf: - {} - type: 'null' title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId title: ToolApprovalRequestedPart description: Tool part in approval-requested state (awaiting user decision). ToolApprovalRespondedPart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: approval-responded title: State default: approval-responded input: anyOf: - {} - type: 'null' title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId title: ToolApprovalRespondedPart description: >- Tool part in approval-responded state (user approved/denied, execution pending). ToolOutputDeniedPart: properties: type: type: string pattern: ^tool- title: Type toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: output-denied title: State default: output-denied input: anyOf: - {} - type: 'null' title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - type - toolCallId title: ToolOutputDeniedPart description: Tool part in output-denied state (tool was denied, terminal state). DynamicToolInputStreamingPart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: input-streaming title: State default: input-streaming input: anyOf: - {} - type: 'null' title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId title: DynamicToolInputStreamingPart description: Dynamic tool part in input-streaming state. DynamicToolInputAvailablePart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: input-available title: State default: input-available input: title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId - input title: DynamicToolInputAvailablePart description: Dynamic tool part in input-available state. DynamicToolApprovalRequestedPart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: approval-requested title: State default: approval-requested input: title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId - input title: DynamicToolApprovalRequestedPart description: Dynamic tool part in approval-requested state (awaiting user decision). DynamicToolApprovalRespondedPart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: approval-responded title: State default: approval-responded input: title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId - input title: DynamicToolApprovalRespondedPart description: >- Dynamic tool part in approval-responded state (user approved/denied, execution pending). DynamicToolOutputDeniedPart: properties: type: type: string const: dynamic-tool title: Type default: dynamic-tool toolName: type: string title: Toolname toolCallId: type: string title: Toolcallid title: anyOf: - type: string - type: 'null' title: Title state: type: string const: output-denied title: State default: output-denied input: title: Input providerExecuted: anyOf: - type: boolean - type: 'null' title: Providerexecuted callProviderMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Callprovidermetadata approval: anyOf: - $ref: '#/components/schemas/ToolApprovalRequested' - $ref: '#/components/schemas/ToolApprovalResponded' - type: 'null' title: Approval additionalProperties: false type: object required: - toolName - toolCallId - input title: DynamicToolOutputDeniedPart description: >- Dynamic tool part in output-denied state (tool was denied, terminal state). SourceUrlUIPart: properties: type: type: string const: source-url title: Type default: source-url sourceId: type: string title: Sourceid url: type: string title: Url title: anyOf: - type: string - type: 'null' title: Title providerMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Providermetadata additionalProperties: false type: object required: - sourceId - url title: SourceUrlUIPart description: A source part of a message. SourceDocumentUIPart: properties: type: type: string const: source-document title: Type default: source-document sourceId: type: string title: Sourceid mediaType: type: string title: Mediatype title: type: string title: Title filename: anyOf: - type: string - type: 'null' title: Filename providerMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Providermetadata additionalProperties: false type: object required: - sourceId - mediaType - title title: SourceDocumentUIPart description: A document source part of a message. FileUIPart: properties: type: type: string const: file title: Type default: file mediaType: type: string title: Mediatype filename: anyOf: - type: string - type: 'null' title: Filename url: type: string title: Url providerMetadata: anyOf: - additionalProperties: additionalProperties: true type: object type: object - type: 'null' title: Providermetadata additionalProperties: false type: object required: - mediaType - url title: FileUIPart description: A file part of a message. DataUIPart: properties: type: type: string pattern: ^data- title: Type id: anyOf: - type: string - type: 'null' title: Id data: title: Data additionalProperties: false type: object required: - type - data title: DataUIPart description: Data part with dynamic type based on data name. StepStartUIPart: properties: type: type: string const: step-start title: Type default: step-start additionalProperties: false type: object title: StepStartUIPart description: A step boundary part of a message. ToolApprovalRequested: properties: id: type: string title: Id additionalProperties: false type: object required: - id title: ToolApprovalRequested description: Tool approval in requested state (awaiting user response). ToolApprovalResponded: properties: id: type: string title: Id approved: type: boolean title: Approved reason: anyOf: - type: string - type: 'null' title: Reason additionalProperties: false type: object required: - id - approved title: ToolApprovalResponded description: Tool approval in responded state (user has approved or denied). PlaygroundExperimentScaffoldContext: properties: name: anyOf: - type: string - type: 'null' title: Name description: anyOf: - type: string - type: 'null' title: Description hasMetadata: type: boolean title: Hasmetadata default: false type: object title: PlaygroundExperimentScaffoldContext description: >- Experiment name/description/metadata the user has staged for the playground's *next* dataset-backed run, before that run has started. The playground UI lets the user pre-set how the next recorded run's experiment will be named, described, and tagged (via the ``set_playground_experiment_recording`` tool or the recording form). That staged state is surfaced here so the agent can see what is already set and avoid re-staging it. Field semantics: - ``name`` / ``description``: the staged values, surfaced to the model verbatim, or ``None`` when the user has not staged them. - ``has_metadata``: a presence flag, not the value. Only *whether* metadata has been staged is model-relevant (so the agent knows not to re-attach it); the metadata object itself is deliberately kept out of the prompt. A field left unstaged (``None`` / ``False``) falls back to the server default when the run starts. The scaffold is consumed once that next run begins. PlaygroundInstanceContext: properties: instanceId: type: integer title: Instanceid model: anyOf: - oneOf: - $ref: '#/components/schemas/PlaygroundBuiltinModelContext' - $ref: '#/components/schemas/PlaygroundCustomProviderModelContext' discriminator: propertyName: type mapping: builtin: $ref: '#/components/schemas/PlaygroundBuiltinModelContext' custom: $ref: '#/components/schemas/PlaygroundCustomProviderModelContext' - type: 'null' title: Model experimentId: anyOf: - type: string - type: 'null' title: Experimentid type: object required: - instanceId title: PlaygroundInstanceContext description: >- One mounted playground instance and its current model selection. ``experiment_id`` carries the relay node id of the experiment produced by this instance's last dataset-backed run, or ``None`` when the instance has not produced one. Ephemeral experiments are included: they remain queryable until the server sweeps them ~24h after their last update. PlaygroundEvaluatorContext: properties: datasetEvaluatorId: type: string title: Datasetevaluatorid name: type: string title: Name kind: type: string enum: - LLM - CODE - BUILTIN title: Kind isBuiltin: type: boolean title: Isbuiltin isApplied: type: boolean title: Isapplied type: object required: - datasetEvaluatorId - name - kind - isBuiltin - isApplied title: PlaygroundEvaluatorContext description: |- One dataset evaluator on the mounted playground's roster. ``name`` is user-controlled; sanitize at every model-visible boundary. ModelProvider: type: string enum: - OPENAI - AZURE_OPENAI - ANTHROPIC - GOOGLE - DEEPSEEK - XAI - OLLAMA - AWS - CEREBRAS - FIREWORKS - GROQ - MOONSHOT - PERPLEXITY - TOGETHER title: ModelProvider PhoenixAssistantMessageMetadata: properties: type: type: string const: assistant title: Type sessionId: type: string title: Sessionid turnTraceContext: anyOf: - $ref: '#/components/schemas/TurnTraceContext' - type: 'null' usage: anyOf: - $ref: '#/components/schemas/AssistantMessageMetadataUsage' - type: 'null' interrupted: type: boolean title: Interrupted default: false additionalProperties: false type: object required: - type - sessionId title: PhoenixAssistantMessageMetadata description: The ``phoenix`` metadata namespace of an assistant message. PhoenixUserMessageMetadata: properties: type: type: string const: user title: Type currentDateTime: type: string maxLength: 128 title: Currentdatetime timeZone: type: string maxLength: 128 title: Timezone isCompactionMessage: type: boolean title: Iscompactionmessage default: false additionalProperties: false type: object required: - type - currentDateTime - timeZone title: PhoenixUserMessageMetadata description: |- The ``phoenix`` metadata namespace the browser attaches to outgoing user messages. PydanticAIMessageMetadata: properties: timestamp: anyOf: - type: string format: date-time - type: 'null' title: Timestamp additionalProperties: false type: object title: PydanticAIMessageMetadata description: |- Local pin of pydantic-ai's message-level ``pydantic_ai`` metadata namespace (its private ``_PydanticAIMessageMetadata``), merged into the assistant metadata by the stream's metadata chunk. PlaygroundBuiltinModelContext: properties: type: type: string const: builtin title: Type default: builtin provider: type: string title: Provider modelName: type: string title: Modelname type: object required: - provider - modelName title: PlaygroundBuiltinModelContext description: Built-in playground model selection. PlaygroundCustomProviderModelContext: properties: type: type: string const: custom title: Type default: custom customProviderId: type: string title: Customproviderid customProviderName: type: string title: Customprovidername provider: type: string title: Provider modelName: type: string title: Modelname type: object required: - customProviderId - customProviderName - provider - modelName title: PlaygroundCustomProviderModelContext description: Custom-provider playground model selection. TurnTraceContext: properties: traceId: type: string pattern: ^[0-9a-f]{32}$ title: Traceid rootSpanId: type: string pattern: ^[0-9a-f]{16}$ title: Rootspanid startedAt: type: string format: date-time title: Startedat additionalProperties: false type: object required: - traceId - rootSpanId - startedAt title: TurnTraceContext description: The trace identity a turn's spans are parented to. AssistantMessageMetadataUsage: properties: tokens: $ref: '#/components/schemas/AssistantMessageMetadataUsageTokens' promptDetails: anyOf: - $ref: >- #/components/schemas/AssistantMessageMetadataUsageCacheTokenDetails - type: 'null' additionalProperties: false type: object required: - tokens title: AssistantMessageMetadataUsage AssistantMessageMetadataUsageTokens: properties: prompt: type: integer title: Prompt completion: type: integer title: Completion total: type: integer title: Total additionalProperties: false type: object required: - prompt - completion - total title: AssistantMessageMetadataUsageTokens AssistantMessageMetadataUsageCacheTokenDetails: properties: cacheRead: type: integer title: Cacheread cacheWrite: type: integer title: Cachewrite additionalProperties: false type: object required: - cacheRead - cacheWrite title: AssistantMessageMetadataUsageCacheTokenDetails description: |- Prompt-cache token counts, mounted as the usage payload's ``prompt_details`` because cached tokens are a breakdown of the prompt. ````