spb/doc-api

1---2title: Events3url: https://platform.claude.com/docs/en/api/beta/sessions/threads/events4---56# Events78## List Session Thread Events910**GET** `/v1/sessions/{session_id}/threads/{thread_id}/events`1112List Session Thread Events1314### Path parameters1516- `session_id: string`1718- `thread_id: string`1920### Query parameters2122- `limit: optional number`2324  Query parameter for limit2526  format: int322728- `page: optional string`2930  Query parameter for page3132### Headers3334- `"anthropic-beta": optional array of AnthropicBeta`3536  Optional header to specify the beta version(s) you want to use.3738  - `string`3940  - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 43 more`4142    - `"message-batches-2024-09-24"`4344    - `"prompt-caching-2024-07-31"`4546    - `"computer-use-2024-10-22"`4748    - `"computer-use-2025-01-24"`4950    - `"pdfs-2024-09-25"`5152    - `"token-counting-2024-11-01"`5354    - `"token-efficient-tools-2025-02-19"`5556    - `"output-128k-2025-02-19"`5758    - `"files-api-2025-04-14"`5960    - `"mcp-client-2025-04-04"`6162    - `"mcp-client-2025-11-20"`6364    - `"dev-full-thinking-2025-05-14"`6566    - `"interleaved-thinking-2025-05-14"`6768    - `"code-execution-2025-05-22"`6970    - `"extended-cache-ttl-2025-04-11"`7172    - `"context-1m-2025-08-07"`7374    - `"context-management-2025-06-27"`7576    - `"model-context-window-exceeded-2025-08-26"`7778    - `"skills-2025-10-02"`7980    - `"fast-mode-2026-02-01"`8182    - `"output-300k-2026-03-24"`8384    - `"user-profiles-2026-03-24"`8586    - `"user-profiles-2026-08-18"`8788    - `"user-profiles-2026-09-04"`8990    - `"advisor-tool-2026-03-01"`9192    - `"managed-agents-2026-04-01"`9394    - `"cache-diagnosis-2026-04-07"`9596    - `"dreaming-2026-04-21"`9798    - `"thinking-token-count-2026-05-13"`99100    - `"server-side-fallback-2026-06-01"`101102    - `"server-side-fallback-2026-07-01"`103104    - `"fallback-credit-2026-06-01"`105106    - `"fallback-credit-2026-07-01"`107108    - `"agent-memory-2026-07-22"`109110    - `"mid-conversation-tool-changes-2026-07-01"`111112    - `"compact-2026-01-12"`113114    - `"computer-use-2025-11-24"`115116    - `"mcp-tunnels-2026-06-22"`117118    - `"structured-outputs-2025-11-13"`119120    - `"task-budgets-2026-03-13"`121122    - `"thinking-display-updates-2026-08-18"`123124    - `"ce-user-management-2026-07-13"`125126    - `"mid-conversation-output-config-2026-07-01"`127128    - `"thinking-binding-controls-2026-08-01"`129130    - `"mid-conversation-system-clear-at-2026-08-21"`131132    - `"compact-2026-09-04"`133134- `"anthropic-workspace-id": optional string`135136### Returns137138- `data: optional array of BetaManagedAgentsSessionEvent`139140  Events for the thread, ordered by `processed_at`.141142  - `BetaManagedAgentsUserMessageEvent object`143144    A user message event in the session conversation.145146    - `type: "user.message"`147148    - `id: string`149150      Unique identifier for this event.151152    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`153154      Array of content blocks comprising the user message.155156      - `BetaManagedAgentsTextBlock object`157158        Regular text content.159160        - `type: "text"`161162        - `text: string`163164          The text content.165166          minLength: 1167168      - `BetaManagedAgentsImageBlock object`169170        Image content specified directly as base64 data or as a reference via a URL.171172        - `type: "image"`173174        - `source: BetaManagedAgentsBase64ImageSource or BetaManagedAgentsURLImageSource or BetaManagedAgentsFileImageSource`175176          Union type for image source variants.177178          - `BetaManagedAgentsBase64ImageSource object`179180            Base64-encoded image data.181182            - `type: "base64"`183184            - `data: string`185186              Base64-encoded image data.187188              minLength: 1189190            - `media_type: string`191192              MIME type of the image (e.g., "image/png", "image/jpeg", "image/gif", "image/webp").193194              minLength: 1195196          - `BetaManagedAgentsURLImageSource object`197198            Image referenced by URL.199200            - `type: "url"`201202            - `url: string`203204              URL of the image to fetch.205206              minLength: 1207208          - `BetaManagedAgentsFileImageSource object`209210            Image referenced by file ID.211212            - `type: "file"`213214            - `file_id: string`215216              ID of a previously uploaded file.217218              minLength: 1219220      - `BetaManagedAgentsDocumentBlock object`221222        Document content, either specified directly as base64 data, as text, or as a reference via a URL.223224        - `type: "document"`225226        - `source: BetaManagedAgentsBase64DocumentSource or BetaManagedAgentsPlainTextDocumentSource or BetaManagedAgentsURLDocumentSource or BetaManagedAgentsFileDocumentSource`227228          Union type for document source variants.229230          - `BetaManagedAgentsBase64DocumentSource object`231232            Base64-encoded document data.233234            - `type: "base64"`235236            - `data: string`237238              Base64-encoded document data.239240              minLength: 1241242            - `media_type: string`243244              MIME type of the document (e.g., "application/pdf").245246              minLength: 1247248          - `BetaManagedAgentsPlainTextDocumentSource object`249250            Plain text document content.251252            - `type: "text"`253254            - `data: string`255256              The plain text content.257258              minLength: 1259260            - `media_type: "text/plain"`261262              MIME type of the text content. Must be "text/plain".263264          - `BetaManagedAgentsURLDocumentSource object`265266            Document referenced by URL.267268            - `type: "url"`269270            - `url: string`271272              URL of the document to fetch.273274              minLength: 1275276          - `BetaManagedAgentsFileDocumentSource object`277278            Document referenced by file ID.279280            - `type: "file"`281282            - `file_id: string`283284              ID of a previously uploaded file.285286              minLength: 1287288        - `context: optional string or null`289290          Additional context about the document for the model.291292        - `title: optional string or null`293294          The title of the document.295296      - `BetaManagedAgentsRedactedBlock object`297298        Placeholder for content withheld by Anthropic model policy.299300        - `type: "redacted"`301302    - `processed_at: optional string or null`303304      A timestamp in RFC 3339 format305306      format: date-time307308  - `BetaManagedAgentsUserInterruptEvent object`309310    An interrupt event that pauses agent execution and returns control to the user.311312    - `type: "user.interrupt"`313314    - `id: string`315316      Unique identifier for this event.317318    - `processed_at: optional string or null`319320      A timestamp in RFC 3339 format321322      format: date-time323324    - `session_thread_id: optional string or null`325326      If absent, interrupts every non-archived thread in a multiagent session (or the primary alone in a single-agent session). If present, interrupts only the named thread.327328  - `BetaManagedAgentsUserToolConfirmationEvent object`329330    A tool confirmation event that approves or denies a pending tool execution.331332    - `type: "user.tool_confirmation"`333334    - `id: string`335336      Unique identifier for this event.337338    - `result: "allow" or "deny"`339340      UserToolConfirmationResult enum341342      - `"allow"`343344      - `"deny"`345346    - `tool_use_id: string`347348      The id of the `agent.tool_use` or `agent.mcp_tool_use` event this result corresponds to, which can be found in the last `session.status_idle` [event's](https://platform.claude.com/docs/en/api/beta/sessions/events/list#beta_managed_agents_session_requires_action.event_ids) `stop_reason.event_ids` field.349350    - `deny_message: optional string or null`351352      Optional message providing context for a 'deny' decision. Only allowed when result is 'deny'.353354      maxLength: 10000355356    - `processed_at: optional string or null`357358      A timestamp in RFC 3339 format359360      format: date-time361362    - `session_thread_id: optional string or null`363364      Set by the server to the subagent thread this confirmation was routed to. Omitted when it was routed to the primary thread.365366  - `BetaManagedAgentsUserCustomToolResultEvent object`367368    Event sent by the client providing the result of a custom tool execution.369370    - `type: "user.custom_tool_result"`371372    - `id: string`373374      Unique identifier for this event.375376    - `custom_tool_use_id: string`377378      The id of the `agent.custom_tool_use` event this result corresponds to, which can be found in the last `session.status_idle` [event's](https://platform.claude.com/docs/en/api/beta/sessions/events/list#beta_managed_agents_session_requires_action.event_ids) `stop_reason.event_ids` field.379380    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`381382      The result content returned by the tool.383384      - `BetaManagedAgentsTextBlock object`385386        Regular text content.387388      - `BetaManagedAgentsImageBlock object`389390        Image content specified directly as base64 data or as a reference via a URL.391392      - `BetaManagedAgentsDocumentBlock object`393394        Document content, either specified directly as base64 data, as text, or as a reference via a URL.395396      - `BetaManagedAgentsSearchResultBlock object`397398        A block containing a web search result.399400        - `type: "search_result"`401402        - `citations: BetaManagedAgentsSearchResultCitations`403404          Citation settings for a search result.405406          - `enabled: boolean`407408            Whether citations are enabled for this search result.409410        - `content: array of BetaManagedAgentsSearchResultContent`411412          Array of text content blocks from the search result.413414          - `type: "text"`415416          - `text: string`417418            The text content.419420            minLength: 1421422        - `source: string`423424          The URL source of the search result.425426          minLength: 1427428        - `title: string`429430          The title of the search result.431432          minLength: 1433434    - `is_error: optional boolean or null`435436      Whether the tool execution resulted in an error.437438    - `processed_at: optional string or null`439440      A timestamp in RFC 3339 format441442      format: date-time443444    - `session_thread_id: optional string or null`445446      Set by the server to the subagent thread this result was routed to. Omitted when it was routed to the primary thread.447448  - `BetaManagedAgentsAgentCustomToolUseEvent object`449450    Event emitted when the agent calls a custom tool. The session goes idle until the client sends a `user.custom_tool_result` event with the result.451452    - `type: "agent.custom_tool_use"`453454    - `id: string`455456      Unique identifier for this event.457458    - `input: map[unknown]`459460      Input parameters for the tool call.461462    - `name: string`463464      Name of the custom tool being called.465466    - `processed_at: string`467468      A timestamp in RFC 3339 format469470      format: date-time471472    - `session_thread_id: optional string or null`473474      When set, this event was cross-posted from a subagent's thread to surface its custom tool use on the primary thread's stream. Empty on the thread's own events. Informational only: the server routes the matching `user.custom_tool_result` by `custom_tool_use_id`, so clients do not send it back.475476  - `BetaManagedAgentsAgentMessageEvent object`477478    An agent response event in the session conversation.479480    - `type: "agent.message"`481482    - `id: string`483484      Unique identifier for this event.485486    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsRedactedBlock`487488      Array of text blocks comprising the agent response.489490      - `BetaManagedAgentsTextBlock object`491492        Regular text content.493494      - `BetaManagedAgentsRedactedBlock object`495496        Placeholder for content withheld by Anthropic model policy.497498    - `processed_at: string`499500      A timestamp in RFC 3339 format501502      format: date-time503504  - `BetaManagedAgentsAgentThinkingEvent object`505506    Indicates the agent is making forward progress via extended thinking. A progress signal, not a content carrier.507508    - `type: "agent.thinking"`509510    - `id: string`511512      Unique identifier for this event.513514    - `processed_at: string`515516      A timestamp in RFC 3339 format517518      format: date-time519520  - `BetaManagedAgentsAgentMCPToolUseEvent object`521522    Event emitted when the agent invokes a tool provided by an MCP server.523524    - `type: "agent.mcp_tool_use"`525526    - `id: string`527528      Unique identifier for this event.529530    - `input: map[unknown]`531532      Input parameters for the tool call.533534    - `mcp_server_name: string`535536      Name of the MCP server providing the tool.537538    - `name: string`539540      Name of the MCP tool being used.541542    - `processed_at: string`543544      A timestamp in RFC 3339 format545546      format: date-time547548    - `evaluated_permission: optional "allow" or "ask" or "deny"`549550      AgentEvaluatedPermission enum551552      - `"allow"`553554      - `"ask"`555556      - `"deny"`557558    - `evaluation: optional BetaManagedAgentsAgentToolEvaluation`559560      Names the resolved permission_policy that produced evaluated_permission, and under auto carries the judgement. Open union: clients must tolerate unknown variants.561562      - `BetaManagedAgentsAgentToolEvaluationAlwaysAllow object`563564        The resolved permission_policy was always_allow; accompanies evaluated_permission "allow".565566        - `type: "always_allow"`567568      - `BetaManagedAgentsAgentToolEvaluationAlwaysAsk object`569570        The resolved permission_policy was always_ask; accompanies evaluated_permission "ask".571572        - `type: "always_ask"`573574      - `BetaManagedAgentsAgentToolEvaluationAuto object`575576        The resolved permission_policy was auto: the server judged this invocation individually.577578        - `type: "auto"`579580        - `evaluated_permission: BetaManagedAgentsAgentAutoEvaluatedPermission`581582          The server's per-invocation judgement under the auto permission policy. Its type always equals the event's top-level evaluated_permission. Open union: clients must tolerate unknown variants.583584          - `BetaManagedAgentsAgentAutoEvaluatedPermissionAllow object`585586            The server judged the invocation safe to execute without client approval.587588            - `type: "allow"`589590          - `BetaManagedAgentsAgentAutoEvaluatedPermissionAsk object`591592            The server reached no judgement; the invocation is held for client approval.593594            - `type: "ask"`595596            - `reason_code: string`597598              The judgement's grounds in registry-bound terms, for client branching and audit rather than end-user display. Open registry; currently "indeterminate" (no judgement was reached). Clients must tolerate values outside this set.599600              maxLength: 64601602          - `BetaManagedAgentsAgentAutoEvaluatedPermissionDeny object`603604            The server judged the invocation high-risk; it does not execute and a synthetic error tool result is appended.605606            - `type: "deny"`607608            - `reason_code: string`609610              The judgement's grounds in registry-bound terms. Open registry; currently "high_risk" (judged high-risk; the call does not run). Clients must tolerate values outside this set.611612              maxLength: 64613614    - `session_thread_id: optional string or null`615616      When set, this event was cross-posted from a subagent's thread to surface its permission request on the primary thread's stream. Empty on the thread's own events. Informational only: the server routes the matching `user.tool_confirmation` by `tool_use_id`, so clients do not send it back.617618  - `BetaManagedAgentsAgentMCPToolResultEvent object`619620    Event representing the result of an MCP tool execution.621622    - `type: "agent.mcp_tool_result"`623624    - `id: string`625626      Unique identifier for this event.627628    - `mcp_tool_use_id: string`629630      The id of the `agent.mcp_tool_use` event this result corresponds to.631632    - `processed_at: string`633634      A timestamp in RFC 3339 format635636      format: date-time637638    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`639640      The result content returned by the tool.641642      - `BetaManagedAgentsTextBlock object`643644        Regular text content.645646      - `BetaManagedAgentsImageBlock object`647648        Image content specified directly as base64 data or as a reference via a URL.649650      - `BetaManagedAgentsDocumentBlock object`651652        Document content, either specified directly as base64 data, as text, or as a reference via a URL.653654      - `BetaManagedAgentsSearchResultBlock object`655656        A block containing a web search result.657658    - `is_error: optional boolean or null`659660      Whether the tool execution resulted in an error.661662  - `BetaManagedAgentsAgentToolUseEvent object`663664    Event emitted when the agent invokes a built-in agent tool.665666    - `type: "agent.tool_use"`667668    - `id: string`669670      Unique identifier for this event.671672    - `input: map[unknown]`673674      Input parameters for the tool call.675676    - `name: string`677678      Name of the agent tool being used.679680    - `processed_at: string`681682      A timestamp in RFC 3339 format683684      format: date-time685686    - `evaluated_permission: optional "allow" or "ask" or "deny"`687688      AgentEvaluatedPermission enum689690      - `"allow"`691692      - `"ask"`693694      - `"deny"`695696    - `evaluation: optional BetaManagedAgentsAgentToolEvaluation`697698      Names the resolved permission_policy that produced evaluated_permission, and under auto carries the judgement. Open union: clients must tolerate unknown variants.699700    - `session_thread_id: optional string or null`701702      When set, this event was cross-posted from a subagent's thread to surface its permission request on the primary thread's stream. Empty on the thread's own events. Informational only: the server routes the matching `user.tool_confirmation` or `user.tool_result` by `tool_use_id`, so clients do not send it back.703704  - `BetaManagedAgentsAgentToolResultEvent object`705706    Event representing the result of an agent tool execution.707708    - `type: "agent.tool_result"`709710    - `id: string`711712      Unique identifier for this event.713714    - `processed_at: string`715716      A timestamp in RFC 3339 format717718      format: date-time719720    - `tool_use_id: string`721722      The id of the `agent.tool_use` event this result corresponds to.723724    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`725726      The result content returned by the tool.727728      - `BetaManagedAgentsTextBlock object`729730        Regular text content.731732      - `BetaManagedAgentsImageBlock object`733734        Image content specified directly as base64 data or as a reference via a URL.735736      - `BetaManagedAgentsDocumentBlock object`737738        Document content, either specified directly as base64 data, as text, or as a reference via a URL.739740      - `BetaManagedAgentsSearchResultBlock object`741742        A block containing a web search result.743744    - `is_error: optional boolean or null`745746      Whether the tool execution resulted in an error.747748  - `BetaManagedAgentsAgentThreadMessageReceivedEvent object`749750    Delivery event written to the target thread's input stream when an agent-to-agent message arrives.751752    - `type: "agent.thread_message_received"`753754    - `id: string`755756      Unique identifier for this event.757758    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`759760      Message content blocks.761762      - `BetaManagedAgentsTextBlock object`763764        Regular text content.765766      - `BetaManagedAgentsImageBlock object`767768        Image content specified directly as base64 data or as a reference via a URL.769770      - `BetaManagedAgentsDocumentBlock object`771772        Document content, either specified directly as base64 data, as text, or as a reference via a URL.773774      - `BetaManagedAgentsRedactedBlock object`775776        Placeholder for content withheld by Anthropic model policy.777778    - `from_session_thread_id: string`779780      Public `sthr_` ID of the thread that sent the message.781782    - `processed_at: string`783784      A timestamp in RFC 3339 format785786      format: date-time787788    - `from_agent_name: optional string or null`789790      Name of the callable agent this message came from. Absent when received from the primary agent.791792  - `BetaManagedAgentsAgentThreadMessageSentEvent object`793794    Observability event emitted to the sender's output stream when an agent-to-agent message is sent.795796    - `type: "agent.thread_message_sent"`797798    - `id: string`799800      Unique identifier for this event.801802    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`803804      Message content blocks.805806      - `BetaManagedAgentsTextBlock object`807808        Regular text content.809810      - `BetaManagedAgentsImageBlock object`811812        Image content specified directly as base64 data or as a reference via a URL.813814      - `BetaManagedAgentsDocumentBlock object`815816        Document content, either specified directly as base64 data, as text, or as a reference via a URL.817818      - `BetaManagedAgentsRedactedBlock object`819820        Placeholder for content withheld by Anthropic model policy.821822    - `processed_at: string`823824      A timestamp in RFC 3339 format825826      format: date-time827828    - `to_session_thread_id: string`829830      Public `sthr_` ID of the thread the message was sent to.831832    - `to_agent_name: optional string or null`833834      Name of the callable agent this message was sent to. Absent when sent to the primary agent.835836  - `BetaManagedAgentsAgentThreadContextCompactedEvent object`837838    Indicates that context compaction (summarization) occurred during the session.839840    - `type: "agent.thread_context_compacted"`841842    - `id: string`843844      Unique identifier for this event.845846    - `processed_at: string`847848      A timestamp in RFC 3339 format849850      format: date-time851852  - `BetaManagedAgentsSessionErrorEvent object`853854    An error event indicating a problem occurred during session execution.855856    - `type: "session.error"`857858    - `id: string`859860      Unique identifier for this event.861862    - `error: BetaManagedAgentsUnknownError or BetaManagedAgentsModelOverloadedError or BetaManagedAgentsModelRateLimitedError or 5 more`863864      - `BetaManagedAgentsUnknownError object`865866        An unknown or unexpected error occurred during session execution. A fallback variant; clients that don't recognize a new error code can match on `retry_status` and `message` alone.867868        - `type: "unknown_error"`869870        - `message: string`871872          Human-readable error description.873874        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`875876          What the client should do next in response to this error.877878          - `BetaManagedAgentsRetryStatusRetrying object`879880            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.881882            - `type: "retrying"`883884          - `BetaManagedAgentsRetryStatusExhausted object`885886            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.887888            - `type: "exhausted"`889890          - `BetaManagedAgentsRetryStatusTerminal object`891892            The session encountered a terminal error and will transition to `terminated` state.893894            - `type: "terminal"`895896      - `BetaManagedAgentsModelOverloadedError object`897898        The model is currently overloaded. Emitted after automatic retries are exhausted.899900        - `type: "model_overloaded_error"`901902        - `message: string`903904          Human-readable error description.905906        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`907908          What the client should do next in response to this error.909910          - `BetaManagedAgentsRetryStatusRetrying object`911912            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.913914          - `BetaManagedAgentsRetryStatusExhausted object`915916            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.917918          - `BetaManagedAgentsRetryStatusTerminal object`919920            The session encountered a terminal error and will transition to `terminated` state.921922      - `BetaManagedAgentsModelRateLimitedError object`923924        The model request was rate-limited.925926        - `type: "model_rate_limited_error"`927928        - `message: string`929930          Human-readable error description.931932        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`933934          What the client should do next in response to this error.935936          - `BetaManagedAgentsRetryStatusRetrying object`937938            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.939940          - `BetaManagedAgentsRetryStatusExhausted object`941942            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.943944          - `BetaManagedAgentsRetryStatusTerminal object`945946            The session encountered a terminal error and will transition to `terminated` state.947948      - `BetaManagedAgentsModelRequestFailedError object`949950        A model request failed for a reason other than overload or rate-limiting.951952        - `type: "model_request_failed_error"`953954        - `message: string`955956          Human-readable error description.957958        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`959960          What the client should do next in response to this error.961962          - `BetaManagedAgentsRetryStatusRetrying object`963964            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.965966          - `BetaManagedAgentsRetryStatusExhausted object`967968            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.969970          - `BetaManagedAgentsRetryStatusTerminal object`971972            The session encountered a terminal error and will transition to `terminated` state.973974      - `BetaManagedAgentsMCPConnectionFailedError object`975976        Failed to connect to an MCP server.977978        - `type: "mcp_connection_failed_error"`979980        - `mcp_server_name: string`981982          Name of the MCP server that failed to connect.983984        - `message: string`985986          Human-readable error description.987988        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`989990          What the client should do next in response to this error.991992          - `BetaManagedAgentsRetryStatusRetrying object`993994            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.995996          - `BetaManagedAgentsRetryStatusExhausted object`997998            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.9991000          - `BetaManagedAgentsRetryStatusTerminal object`10011002            The session encountered a terminal error and will transition to `terminated` state.10031004      - `BetaManagedAgentsMCPAuthenticationFailedError object`10051006        Authentication to an MCP server failed.10071008        - `type: "mcp_authentication_failed_error"`10091010        - `mcp_server_name: string`10111012          Name of the MCP server that failed authentication.10131014        - `message: string`10151016          Human-readable error description.10171018        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`10191020          What the client should do next in response to this error.10211022          - `BetaManagedAgentsRetryStatusRetrying object`10231024            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.10251026          - `BetaManagedAgentsRetryStatusExhausted object`10271028            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.10291030          - `BetaManagedAgentsRetryStatusTerminal object`10311032            The session encountered a terminal error and will transition to `terminated` state.10331034      - `BetaManagedAgentsBillingError object`10351036        The caller's organization or workspace cannot make model requests — out of credits or spend limit reached. Retrying with the same credentials will not succeed; the caller must resolve the billing state.10371038        - `type: "billing_error"`10391040        - `message: string`10411042          Human-readable error description.10431044        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`10451046          What the client should do next in response to this error.10471048          - `BetaManagedAgentsRetryStatusRetrying object`10491050            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.10511052          - `BetaManagedAgentsRetryStatusExhausted object`10531054            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.10551056          - `BetaManagedAgentsRetryStatusTerminal object`10571058            The session encountered a terminal error and will transition to `terminated` state.10591060      - `BetaManagedAgentsCredentialHostUnreachableError object`10611062        An `environment_variable` credential's `auth.networking.allowed_hosts` includes a host the environment's network policy does not permit.10631064        - `type: "credential_host_unreachable_error"`10651066        - `credential_id: string`10671068          ID of the affected credential.10691070        - `message: string`10711072          Human-readable error description.10731074        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`10751076          What the client should do next in response to this error.10771078          - `BetaManagedAgentsRetryStatusRetrying object`10791080            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.10811082          - `BetaManagedAgentsRetryStatusExhausted object`10831084            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.10851086          - `BetaManagedAgentsRetryStatusTerminal object`10871088            The session encountered a terminal error and will transition to `terminated` state.10891090        - `vault_id: string`10911092          ID of the vault containing the affected credential.10931094    - `processed_at: string`10951096      A timestamp in RFC 3339 format10971098      format: date-time10991100  - `BetaManagedAgentsSessionStatusRescheduledEvent object`11011102    Indicates the session is recovering from an error state and is rescheduled for execution.11031104    - `type: "session.status_rescheduled"`11051106    - `id: string`11071108      Unique identifier for this event.11091110    - `processed_at: string`11111112      A timestamp in RFC 3339 format11131114      format: date-time11151116  - `BetaManagedAgentsSessionStatusRunningEvent object`11171118    Indicates the session is actively running and the agent is working.11191120    - `type: "session.status_running"`11211122    - `id: string`11231124      Unique identifier for this event.11251126    - `processed_at: string`11271128      A timestamp in RFC 3339 format11291130      format: date-time11311132  - `BetaManagedAgentsSessionStatusIdleEvent object`11331134    Indicates the agent has paused and is awaiting user input.11351136    - `type: "session.status_idle"`11371138    - `id: string`11391140      Unique identifier for this event.11411142    - `processed_at: string`11431144      A timestamp in RFC 3339 format11451146      format: date-time11471148    - `stop_reason: BetaManagedAgentsSessionEndTurn or BetaManagedAgentsSessionRequiresAction or BetaManagedAgentsSessionRetriesExhausted or BetaManagedAgentsSessionBudgetReached`11491150      - `BetaManagedAgentsSessionEndTurn object`11511152        The agent completed its turn naturally and is ready for the next user message.11531154        - `type: "end_turn"`11551156      - `BetaManagedAgentsSessionRequiresAction object`11571158        The agent is idle waiting on one or more blocking user-input events (tool confirmation, custom tool result, etc.). Resolving all of them transitions the session back to running.11591160        - `type: "requires_action"`11611162        - `event_ids: array of string`11631164          The ids of events the agent is blocked on. Resolving fewer than all re-emits `session.status_idle` with the remainder.11651166      - `BetaManagedAgentsSessionRetriesExhausted object`11671168        The turn ended because repeated errors exhausted the retry budget or an error escalated to `retry_status: 'exhausted'`.11691170        - `type: "retries_exhausted"`11711172      - `BetaManagedAgentsSessionBudgetReached object`11731174        The agent stopped because the session's tracked list cost reached its budget, or because its usage includes a model with no list price (which the budget cannot measure). Raise the budget to continue — or, if raising is rejected because a model has no list price, remove the budget.11751176        - `type: "budget_reached"`11771178  - `BetaManagedAgentsSessionStatusTerminatedEvent object`11791180    Indicates the session has terminated, either due to an error or completion.11811182    - `type: "session.status_terminated"`11831184    - `id: string`11851186      Unique identifier for this event.11871188    - `processed_at: string`11891190      A timestamp in RFC 3339 format11911192      format: date-time11931194  - `BetaManagedAgentsSessionThreadCreatedEvent object`11951196    Emitted when a subagent is spawned as a new thread. Written to the parent thread's output stream so clients observing the session see child creation.11971198    - `type: "session.thread_created"`11991200    - `id: string`12011202      Unique identifier for this event.12031204    - `agent_name: string`12051206      Name of the callable agent the thread runs.12071208    - `processed_at: string`12091210      A timestamp in RFC 3339 format12111212      format: date-time12131214    - `session_thread_id: string`12151216      Public `sthr_` ID of the newly created thread.12171218  - `BetaManagedAgentsSpanOutcomeEvaluationStartEvent object`12191220    Emitted when an outcome evaluation cycle begins.12211222    - `type: "span.outcome_evaluation_start"`12231224    - `id: string`12251226      Unique identifier for this event.12271228    - `iteration: number`12291230      0-indexed revision cycle. 0 is the first evaluation; 1 is the re-evaluation after the first revision; etc.12311232      format: int3212331234    - `outcome_id: string`12351236      The `outc_` ID of the outcome being evaluated.12371238    - `processed_at: string`12391240      A timestamp in RFC 3339 format12411242      format: date-time12431244  - `BetaManagedAgentsSpanOutcomeEvaluationEndEvent object`12451246    Emitted when an outcome evaluation cycle completes. Carries the verdict and aggregate token usage. A verdict of `needs_revision` means another evaluation cycle follows; `satisfied`, `max_iterations_reached`, `failed`, or `interrupted` are terminal — no further evaluation cycles follow.12471248    - `type: "span.outcome_evaluation_end"`12491250    - `id: string`12511252      Unique identifier for this event.12531254    - `explanation: string`12551256      Human-readable explanation of the verdict. For `needs_revision`, describes which criteria failed and why.12571258    - `iteration: number`12591260      0-indexed revision cycle, matching the corresponding `span.outcome_evaluation_start`.12611262      format: int3212631264    - `outcome_evaluation_start_id: string`12651266      The id of the corresponding `span.outcome_evaluation_start` event.12671268    - `outcome_id: string`12691270      The `outc_` ID of the outcome being evaluated.12711272    - `processed_at: string`12731274      A timestamp in RFC 3339 format12751276      format: date-time12771278    - `result: string`12791280      Evaluation verdict. 'satisfied': criteria met, session goes idle. 'needs_revision': criteria not met, another revision cycle follows. 'max_iterations_reached': evaluation budget exhausted with criteria still unmet — one final acknowledgment turn follows before the session goes idle, but no further evaluation runs. 'failed': grader determined the rubric does not apply to the deliverables. 'interrupted': user sent an interrupt while evaluation was in progress.12811282    - `usage: BetaManagedAgentsSpanModelUsage`12831284      Token usage for a single model request.12851286      - `cache_creation_input_tokens: number`12871288        Tokens used to create prompt cache in this request.12891290        format: int3212911292      - `cache_read_input_tokens: number`12931294        Tokens read from prompt cache in this request.12951296        format: int3212971298      - `input_tokens: number`12991300        Input tokens consumed by this request.13011302        format: int3213031304      - `output_tokens: number`13051306        Output tokens generated by this request.13071308        format: int3213091310      - `speed: optional "standard" or "fast" or null`13111312        Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Not all models support `fast`; invalid combinations are rejected at create time.13131314        - `"standard"`13151316        - `"fast"`13171318  - `BetaManagedAgentsSpanModelRequestStartEvent object`13191320    Emitted when a model request is initiated by the agent.13211322    - `type: "span.model_request_start"`13231324    - `id: string`13251326      Unique identifier for this event.13271328    - `processed_at: string`13291330      A timestamp in RFC 3339 format13311332      format: date-time13331334  - `BetaManagedAgentsSpanModelRequestEndEvent object`13351336    Emitted when a model request completes.13371338    - `type: "span.model_request_end"`13391340    - `id: string`13411342      Unique identifier for this event.13431344    - `is_error: boolean or null`13451346      Whether the model request resulted in an error.13471348    - `model_request_start_id: string`13491350      The id of the corresponding `span.model_request_start` event.13511352    - `model_usage: BetaManagedAgentsSpanModelUsage`13531354      Token usage for a single model request.13551356    - `processed_at: string`13571358      A timestamp in RFC 3339 format13591360      format: date-time13611362  - `BetaManagedAgentsSpanOutcomeEvaluationOngoingEvent object`13631364    Periodic heartbeat emitted while an outcome evaluation cycle is in progress. Distinguishes 'evaluation is actively running' from 'evaluation is stuck' between the corresponding `span.outcome_evaluation_start` and `span.outcome_evaluation_end` events.13651366    - `type: "span.outcome_evaluation_ongoing"`13671368    - `id: string`13691370      Unique identifier for this event.13711372    - `iteration: number`13731374      0-indexed revision cycle, matching the corresponding `span.outcome_evaluation_start`.13751376      format: int3213771378    - `outcome_id: string`13791380      The `outc_` ID of the outcome being evaluated.13811382    - `processed_at: string`13831384      A timestamp in RFC 3339 format13851386      format: date-time13871388  - `BetaManagedAgentsUserDefineOutcomeEvent object`13891390    Echo of a `user.define_outcome` input event. Carries the server-generated `outcome_id` that subsequent `span.outcome_evaluation_*` events reference.13911392    - `type: "user.define_outcome"`13931394    - `id: string`13951396      Unique identifier for this event.13971398    - `description: string`13991400      What the agent should produce. Copied from the input event.14011402    - `max_iterations: number or null`14031404      Evaluate-then-revise cycles before giving up. Default 3, max 20.14051406      format: int3214071408    - `outcome_id: string`14091410      Server-generated `outc_` ID for this outcome. Referenced by `span.outcome_evaluation_*` events and the session's `outcome_evaluations` list.14111412    - `processed_at: string`14131414      A timestamp in RFC 3339 format14151416      format: date-time14171418    - `rubric: BetaManagedAgentsFileRubric or BetaManagedAgentsTextRubric`14191420      Rubric for grading the quality of an outcome.14211422      - `BetaManagedAgentsFileRubric object`14231424        Rubric referenced by a file uploaded via the Files API.14251426        - `type: "file"`14271428        - `file_id: string`14291430          ID of the rubric file.14311432      - `BetaManagedAgentsTextRubric object`14331434        Rubric content provided inline as text.14351436        - `type: "text"`14371438        - `content: string`14391440          Rubric content. Plain text or markdown — the grader treats it as freeform text.14411442  - `BetaManagedAgentsSessionDeletedEvent object`14431444    Emitted when a session has been deleted. Terminates any active event stream — no further events will be emitted for this session.14451446    - `type: "session.deleted"`14471448    - `id: string`14491450      Unique identifier for this event.14511452    - `processed_at: string`14531454      A timestamp in RFC 3339 format14551456      format: date-time14571458  - `BetaManagedAgentsSessionThreadStatusRunningEvent object`14591460    A session thread has begun executing. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.14611462    - `type: "session.thread_status_running"`14631464    - `id: string`14651466      Unique identifier for this event.14671468    - `agent_name: string`14691470      Name of the agent the thread runs.14711472    - `processed_at: string`14731474      A timestamp in RFC 3339 format14751476      format: date-time14771478    - `session_thread_id: string`14791480      Public sthr_ ID of the thread that started running.14811482  - `BetaManagedAgentsSessionThreadStatusIdleEvent object`14831484    A session thread has yielded and is awaiting input. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.14851486    - `type: "session.thread_status_idle"`14871488    - `id: string`14891490      Unique identifier for this event.14911492    - `agent_name: string`14931494      Name of the agent the thread runs.14951496    - `processed_at: string`14971498      A timestamp in RFC 3339 format14991500      format: date-time15011502    - `session_thread_id: string`15031504      Public sthr_ ID of the thread that went idle.15051506    - `stop_reason: BetaManagedAgentsSessionEndTurn or BetaManagedAgentsSessionRequiresAction or BetaManagedAgentsSessionRetriesExhausted or BetaManagedAgentsSessionBudgetReached`15071508      - `BetaManagedAgentsSessionEndTurn object`15091510        The agent completed its turn naturally and is ready for the next user message.15111512      - `BetaManagedAgentsSessionRequiresAction object`15131514        The agent is idle waiting on one or more blocking user-input events (tool confirmation, custom tool result, etc.). Resolving all of them transitions the session back to running.15151516      - `BetaManagedAgentsSessionRetriesExhausted object`15171518        The turn ended because repeated errors exhausted the retry budget or an error escalated to `retry_status: 'exhausted'`.15191520      - `BetaManagedAgentsSessionBudgetReached object`15211522        The agent stopped because the session's tracked list cost reached its budget, or because its usage includes a model with no list price (which the budget cannot measure). Raise the budget to continue — or, if raising is rejected because a model has no list price, remove the budget.15231524  - `BetaManagedAgentsSessionThreadStatusTerminatedEvent object`15251526    A session thread has terminated and will accept no further input. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.15271528    - `type: "session.thread_status_terminated"`15291530    - `id: string`15311532      Unique identifier for this event.15331534    - `agent_name: string`15351536      Name of the agent the thread runs.15371538    - `processed_at: string`15391540      A timestamp in RFC 3339 format15411542      format: date-time15431544    - `session_thread_id: string`15451546      Public sthr_ ID of the thread that terminated.15471548  - `BetaManagedAgentsUserToolResultEvent object`15491550    Event sent by the client providing the result of an agent-toolset tool execution. Only valid on `self_hosted` environments, where sandbox-routed tools are executed by the client rather than the server.15511552    - `type: "user.tool_result"`15531554    - `id: string`15551556      Unique identifier for this event.15571558    - `tool_use_id: string`15591560      The id of the `agent.tool_use` event this result corresponds to, which can be found in the last `session.status_idle` [event's](https://platform.claude.com/docs/en/api/beta/sessions/events/list#beta_managed_agents_session_requires_action.event_ids) `stop_reason.event_ids` field.15611562    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`15631564      The result content returned by the tool.15651566      - `BetaManagedAgentsTextBlock object`15671568        Regular text content.15691570      - `BetaManagedAgentsImageBlock object`15711572        Image content specified directly as base64 data or as a reference via a URL.15731574      - `BetaManagedAgentsDocumentBlock object`15751576        Document content, either specified directly as base64 data, as text, or as a reference via a URL.15771578      - `BetaManagedAgentsSearchResultBlock object`15791580        A block containing a web search result.15811582    - `is_error: optional boolean or null`15831584      Whether the tool execution resulted in an error.15851586    - `processed_at: optional string or null`15871588      A timestamp in RFC 3339 format15891590      format: date-time15911592    - `session_thread_id: optional string or null`15931594      Set by the server to the subagent thread this result was routed to. Omitted when it was routed to the primary thread.15951596  - `BetaManagedAgentsSessionThreadStatusRescheduledEvent object`15971598    A session thread hit a transient error and is retrying automatically. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.15991600    - `type: "session.thread_status_rescheduled"`16011602    - `id: string`16031604      Unique identifier for this event.16051606    - `agent_name: string`16071608      Name of the agent the thread runs.16091610    - `processed_at: string`16111612      A timestamp in RFC 3339 format16131614      format: date-time16151616    - `session_thread_id: string`16171618      Public sthr_ ID of the thread that is retrying.16191620  - `BetaManagedAgentsSessionUpdatedEvent object`16211622    Emitted when an UpdateSession request changed at least one field. Carries only the fields that changed; absent fields were not part of the update. The new configuration applies from the next turn.16231624    - `type: "session.updated"`16251626    - `id: string`16271628      Unique identifier for this event.16291630    - `processed_at: string`16311632      A timestamp in RFC 3339 format16331634      format: date-time16351636    - `agent: optional BetaManagedAgentsSessionAgent or null`16371638      Resolved `agent` definition for a `session`. Snapshot of the `agent` at `session` creation time.16391640      - `type: "agent"`16411642      - `id: string`16431644      - `description: string or null`16451646      - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`16471648        - `type: "url"`16491650        - `name: string`16511652        - `url: string`16531654      - `model: BetaManagedAgentsModelConfig`16551656        Model identifier and configuration.16571658        - `id: BetaManagedAgentsModel`16591660          The model that will power your agent.16611662          See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.16631664          - `"claude-fable-5-1" or "claude-sonnet-5" or "claude-fable-5" or 11 more`16651666            The model that will power your agent.16671668            See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.16691670            - `"claude-fable-5-1"`16711672              Frontier intelligence for ambitious tasks across coding, scientific discovery, and enterprise workflows16731674            - `"claude-sonnet-5"`16751676              High-performance model for coding and agents16771678            - `"claude-fable-5"`16791680              Next generation of intelligence for the hardest knowledge work and coding problems16811682            - `"claude-opus-5"`16831684              Powerful intelligence for long-running agents and coding16851686            - `"claude-opus-4-8"`16871688              Powerful intelligence for long-running agents and coding16891690            - `"claude-opus-4-7"`16911692              Powerful intelligence for long-running agents and coding16931694            - `"claude-opus-4-6"`16951696              Powerful intelligence for long-running agents and coding16971698            - `"claude-sonnet-4-6"`16991700              Best combination of speed and intelligence17011702            - `"claude-haiku-4-5"`17031704              Fastest model with near-frontier intelligence17051706            - `"claude-haiku-4-5-20251001"`17071708              Fastest model with near-frontier intelligence17091710            - `"claude-opus-4-5"`17111712              Powerful intelligence for long-running agents and coding17131714            - `"claude-opus-4-5-20251101"`17151716              Powerful intelligence for long-running agents and coding17171718            - `"claude-sonnet-4-5"`17191720              High-performance model for agents and coding17211722            - `"claude-sonnet-4-5-20250929"`17231724              High-performance model for agents and coding17251726          - `string`17271728        - `effort: optional BetaManagedAgentsEffortLow or BetaManagedAgentsEffortMedium or BetaManagedAgentsEffortHigh or 2 more`17291730          How hard Claude works on each turn. Sets `output_config.effort` on every Messages call the session makes.17311732          - `BetaManagedAgentsEffortLow object`17331734            Low effort. Favors latency over reasoning depth.17351736            - `type: "low"`17371738          - `BetaManagedAgentsEffortMedium object`17391740            Medium effort. Balances latency and reasoning depth.17411742            - `type: "medium"`17431744          - `BetaManagedAgentsEffortHigh object`17451746            High effort. Favors reasoning depth.17471748            - `type: "high"`17491750          - `BetaManagedAgentsEffortXhigh object`17511752            Extra-high effort. Not all models accept this level.17531754            - `type: "xhigh"`17551756          - `BetaManagedAgentsEffortMax object`17571758            Maximum effort. Favors reasoning depth over latency.17591760            - `type: "max"`17611762        - `inference_geo: optional string`17631764          Geographic region for model inference. When unset, requests fall through to the workspace's default_inference_geo.17651766        - `speed: optional "standard" or "fast"`17671768          Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Not all models support `fast`; invalid combinations are rejected at create time.17691770          - `"standard"`17711772          - `"fast"`17731774      - `multiagent: BetaManagedAgentsSessionMultiagentCoordinator or null`17751776        Resolved coordinator topology with full agent definitions for each roster member.17771778        - `type: "coordinator"`17791780        - `agents: array of BetaManagedAgentsSessionThreadAgent or BetaManagedAgentsAdvisor`17811782          Full `agent` definitions the coordinator may spawn as session threads.17831784          - `BetaManagedAgentsSessionThreadAgent object`17851786            Resolved `agent` definition for a single `session_thread`. Snapshot of the agent at thread creation time. The multiagent roster is not repeated here; read it from `Session.agent`.17871788            - `type: "agent"`17891790            - `id: string`17911792            - `description: string or null`17931794            - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`17951796              - `type: "url"`17971798              - `name: string`17991800              - `url: string`18011802            - `model: BetaManagedAgentsModelConfig`18031804              Model identifier and configuration.18051806            - `name: string`18071808            - `skills: array of BetaManagedAgentsAnthropicSkill or BetaManagedAgentsCustomSkill`18091810              - `BetaManagedAgentsAnthropicSkill object`18111812                A resolved Anthropic-managed skill.18131814                - `type: "anthropic"`18151816                - `skill_id: string`18171818                - `version: string`18191820              - `BetaManagedAgentsCustomSkill object`18211822                A resolved user-created custom skill.18231824                - `type: "custom"`18251826                - `skill_id: string`18271828                - `version: string`18291830            - `system: string or null`18311832            - `tools: array of BetaManagedAgentsAgentToolset20260401 or BetaManagedAgentsMCPToolset or BetaManagedAgentsCustomTool`18331834              - `BetaManagedAgentsAgentToolset20260401 object`18351836                - `type: "agent_toolset_20260401"`18371838                - `configs: array of BetaManagedAgentsAgentToolConfig`18391840                  - `BetaManagedAgentsBashToolConfig object`18411842                    Configuration for the bash tool.18431844                    - `type: "bash"`18451846                    - `enabled: boolean`18471848                    - `name: "bash"`18491850                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`18511852                      Permission policy for tool execution.18531854                      - `BetaManagedAgentsAlwaysAllowPolicy object`18551856                        Tool calls are automatically approved without user confirmation.18571858                        - `type: "always_allow"`18591860                      - `BetaManagedAgentsAlwaysAskPolicy object`18611862                        Tool calls require user confirmation before execution.18631864                        - `type: "always_ask"`18651866                      - `BetaManagedAgentsAutoPolicy object`18671868                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.18691870                        - `type: "auto"`18711872                  - `BetaManagedAgentsEditToolConfig object`18731874                    Configuration for the edit tool.18751876                    - `type: "edit"`18771878                    - `enabled: boolean`18791880                    - `name: "edit"`18811882                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`18831884                      Permission policy for tool execution.18851886                      - `BetaManagedAgentsAlwaysAllowPolicy object`18871888                        Tool calls are automatically approved without user confirmation.18891890                      - `BetaManagedAgentsAlwaysAskPolicy object`18911892                        Tool calls require user confirmation before execution.18931894                      - `BetaManagedAgentsAutoPolicy object`18951896                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.18971898                  - `BetaManagedAgentsReadToolConfig object`18991900                    Configuration for the read tool.19011902                    - `type: "read"`19031904                    - `enabled: boolean`19051906                    - `name: "read"`19071908                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`19091910                      Permission policy for tool execution.19111912                      - `BetaManagedAgentsAlwaysAllowPolicy object`19131914                        Tool calls are automatically approved without user confirmation.19151916                      - `BetaManagedAgentsAlwaysAskPolicy object`19171918                        Tool calls require user confirmation before execution.19191920                      - `BetaManagedAgentsAutoPolicy object`19211922                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.19231924                  - `BetaManagedAgentsWriteToolConfig object`19251926                    Configuration for the write tool.19271928                    - `type: "write"`19291930                    - `enabled: boolean`19311932                    - `name: "write"`19331934                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`19351936                      Permission policy for tool execution.19371938                      - `BetaManagedAgentsAlwaysAllowPolicy object`19391940                        Tool calls are automatically approved without user confirmation.19411942                      - `BetaManagedAgentsAlwaysAskPolicy object`19431944                        Tool calls require user confirmation before execution.19451946                      - `BetaManagedAgentsAutoPolicy object`19471948                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.19491950                  - `BetaManagedAgentsGlobToolConfig object`19511952                    Configuration for the glob tool.19531954                    - `type: "glob"`19551956                    - `enabled: boolean`19571958                    - `name: "glob"`19591960                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`19611962                      Permission policy for tool execution.19631964                      - `BetaManagedAgentsAlwaysAllowPolicy object`19651966                        Tool calls are automatically approved without user confirmation.19671968                      - `BetaManagedAgentsAlwaysAskPolicy object`19691970                        Tool calls require user confirmation before execution.19711972                      - `BetaManagedAgentsAutoPolicy object`19731974                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.19751976                  - `BetaManagedAgentsGrepToolConfig object`19771978                    Configuration for the grep tool.19791980                    - `type: "grep"`19811982                    - `enabled: boolean`19831984                    - `name: "grep"`19851986                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`19871988                      Permission policy for tool execution.19891990                      - `BetaManagedAgentsAlwaysAllowPolicy object`19911992                        Tool calls are automatically approved without user confirmation.19931994                      - `BetaManagedAgentsAlwaysAskPolicy object`19951996                        Tool calls require user confirmation before execution.19971998                      - `BetaManagedAgentsAutoPolicy object`19992000                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.20012002                  - `BetaManagedAgentsWebFetchToolConfig object`20032004                    Configuration for the web_fetch tool.20052006                    - `type: "web_fetch"`20072008                    - `enabled: boolean`20092010                    - `name: "web_fetch"`20112012                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`20132014                      Permission policy for tool execution.20152016                      - `BetaManagedAgentsAlwaysAllowPolicy object`20172018                        Tool calls are automatically approved without user confirmation.20192020                      - `BetaManagedAgentsAlwaysAskPolicy object`20212022                        Tool calls require user confirmation before execution.20232024                      - `BetaManagedAgentsAutoPolicy object`20252026                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.20272028                    - `allowed_domains: optional array of string`20292030                    - `blocked_domains: optional array of string`20312032                    - `max_content_tokens: optional number or null`20332034                      format: int3220352036                  - `BetaManagedAgentsWebSearchToolConfig object`20372038                    Configuration for the web_search tool.20392040                    - `type: "web_search"`20412042                    - `enabled: boolean`20432044                    - `name: "web_search"`20452046                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`20472048                      Permission policy for tool execution.20492050                      - `BetaManagedAgentsAlwaysAllowPolicy object`20512052                        Tool calls are automatically approved without user confirmation.20532054                      - `BetaManagedAgentsAlwaysAskPolicy object`20552056                        Tool calls require user confirmation before execution.20572058                      - `BetaManagedAgentsAutoPolicy object`20592060                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.20612062                    - `allowed_domains: optional array of string`20632064                    - `blocked_domains: optional array of string`20652066                    - `user_location: optional BetaManagedAgentsUserLocation or null`20672068                      Approximate user location for search result localization.20692070                      - `type: "approximate"`20712072                        Location precision. Only "approximate" is supported.20732074                      - `city: optional string or null`20752076                        City name.20772078                        minLength: 1, maxLength: 25520792080                      - `country: optional string or null`20812082                        Two-letter ISO 3166-1 country code, uppercase.20832084                      - `region: optional string or null`20852086                        Region or state name.20872088                        minLength: 1, maxLength: 25520892090                      - `timezone: optional string or null`20912092                        IANA timezone identifier, e.g. "America/Los_Angeles".20932094                        minLength: 1, maxLength: 25520952096                - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`20972098                  Resolved default configuration for agent tools.20992100                  - `enabled: boolean`21012102                  - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`21032104                    Permission policy for tool execution.21052106                    - `BetaManagedAgentsAlwaysAllowPolicy object`21072108                      Tool calls are automatically approved without user confirmation.21092110                    - `BetaManagedAgentsAlwaysAskPolicy object`21112112                      Tool calls require user confirmation before execution.21132114                    - `BetaManagedAgentsAutoPolicy object`21152116                      The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.21172118              - `BetaManagedAgentsMCPToolset object`21192120                - `type: "mcp_toolset"`21212122                - `configs: array of BetaManagedAgentsMCPToolConfig`21232124                  - `enabled: boolean`21252126                  - `name: string`21272128                  - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`21292130                    Permission policy for tool execution.21312132                    - `BetaManagedAgentsAlwaysAllowPolicy object`21332134                      Tool calls are automatically approved without user confirmation.21352136                    - `BetaManagedAgentsAlwaysAskPolicy object`21372138                      Tool calls require user confirmation before execution.21392140                    - `BetaManagedAgentsAutoPolicy object`21412142                      The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.21432144                - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`21452146                  Resolved default configuration for all tools from an MCP server.21472148                  - `enabled: boolean`21492150                  - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`21512152                    Permission policy for tool execution.21532154                    - `BetaManagedAgentsAlwaysAllowPolicy object`21552156                      Tool calls are automatically approved without user confirmation.21572158                    - `BetaManagedAgentsAlwaysAskPolicy object`21592160                      Tool calls require user confirmation before execution.21612162                    - `BetaManagedAgentsAutoPolicy object`21632164                      The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.21652166                - `mcp_server_name: string`21672168              - `BetaManagedAgentsCustomTool object`21692170                A custom tool as returned in API responses.21712172                - `type: "custom"`21732174                - `description: string`21752176                - `input_schema: BetaManagedAgentsCustomToolInputSchema`21772178                  JSON Schema for custom tool input parameters.21792180                  - `type: "object"`21812182                  - `properties: optional map[unknown] or null`21832184                  - `required: optional array of string or null`21852186                - `name: string`21872188            - `version: number`21892190              format: int3221912192          - `BetaManagedAgentsAdvisor object`21932194            Platform advisor roster entry: a model the session's primary thread may consult mid-turn.21952196            - `type: "advisor"`21972198            - `model: string`21992200              The advisor model id.22012202      - `name: string`22032204      - `skills: array of BetaManagedAgentsAnthropicSkill or BetaManagedAgentsCustomSkill`22052206        - `BetaManagedAgentsAnthropicSkill object`22072208          A resolved Anthropic-managed skill.22092210        - `BetaManagedAgentsCustomSkill object`22112212          A resolved user-created custom skill.22132214      - `system: string or null`22152216      - `tools: array of BetaManagedAgentsAgentToolset20260401 or BetaManagedAgentsMCPToolset or BetaManagedAgentsCustomTool`22172218        - `BetaManagedAgentsAgentToolset20260401 object`22192220        - `BetaManagedAgentsMCPToolset object`22212222        - `BetaManagedAgentsCustomTool object`22232224          A custom tool as returned in API responses.22252226      - `version: number`22272228        format: int3222292230    - `budget: optional BetaManagedAgentsBudgetLimit or null`22312232      A hard spend ceiling. The session stops issuing new model requests once the tracked list cost reaches `max_list_cost`.22332234      - `type: "limit"`22352236      - `max_list_cost: BetaMonetaryAmount`22372238        A monetary amount in a specific currency.22392240        - `amount: string`22412242          Amount in minor units of the currency, as an integer decimal string with no leading zeros: "2500" is $25.00 and "50" is fifty cents. A string rather than a number so no float rounding is ever applied.22432244        - `currency: BetaCurrency`22452246          Uppercase ISO-4217 currency code. `USD` is the only currency currently supported; the accepted set is closed and grows only when a new currency is priced.22472248    - `metadata: optional map[string]`22492250      The session's full metadata bag after the update. Present when the update set non-empty metadata; absent when metadata was unchanged or cleared to empty.22512252    - `title: optional string or null`22532254      The session's new title. Present only when the update changed it.22552256  - `BetaManagedAgentsSystemMessageEvent object`22572258    A mid-conversation system message event. Carries system-role content that is appended to the session as a `role: "system"` turn.22592260    - `type: "system.message"`22612262    - `id: string`22632264      Unique identifier for this event.22652266    - `content: array of BetaManagedAgentsSystemContentBlock`22672268      System content blocks. Text-only.22692270      - `type: "text"`22712272      - `text: string`22732274        The text content.22752276        minLength: 122772278    - `processed_at: optional string or null`22792280      A timestamp in RFC 3339 format22812282      format: date-time22832284  - `BetaManagedAgentsSessionUsageEvent object`22852286    Periodic snapshot of the session's cumulative usage and tracked list cost.22872288    - `type: "session.usage"`22892290    - `id: string`22912292      Unique identifier for this event.22932294    - `processed_at: string`22952296      A timestamp in RFC 3339 format22972298      format: date-time22992300    - `usage: BetaManagedAgentsSessionUsageSnapshot`23012302      Point-in-time snapshot of a session's cumulative usage.23032304      - `active_seconds: optional number`23052306        Cumulative time in seconds during which the session had at least one thread in running status. Overlapping activity from concurrent threads is counted once. This is the duration the session's runtime cost is priced on.23072308        format: double23092310      - `cache_creation: optional BetaManagedAgentsCacheCreationUsage`23112312        Prompt-cache creation token usage broken down by cache lifetime.23132314        - `ephemeral_1h_input_tokens: optional number`23152316          Tokens used to create 1-hour ephemeral cache entries.23172318          format: int3223192320        - `ephemeral_5m_input_tokens: optional number`23212322          Tokens used to create 5-minute ephemeral cache entries.23232324          format: int3223252326      - `cache_read_input_tokens: optional number`23272328        Total tokens read from prompt cache.23292330        format: int3223312332      - `input_tokens: optional number`23332334        Total input tokens consumed across all turns.23352336        format: int3223372338      - `list_cost: optional BetaMonetaryAmount`23392340        A monetary amount in a specific currency.23412342      - `output_tokens: optional number`23432344        Total output tokens generated across all turns.23452346        format: int3223472348      - `server_tool_use: optional BetaManagedAgentsServerToolUsage`23492350        Cumulative count of server-executed tool invocations, broken down by tool.23512352        - `web_fetch_requests: optional number`23532354          Number of server-executed web fetch requests.23552356          format: int3223572358        - `web_search_requests: optional number`23592360          Number of server-executed web search requests.23612362          format: int3223632364    - `budget: optional BetaManagedAgentsBudgetLimit or null`23652366      A hard spend ceiling. The session stops issuing new model requests once the tracked list cost reaches `max_list_cost`.23672368- `next_page: optional string or null`23692370  Opaque cursor for the next page. Null when no more results.23712372### Example23732374```bash2375curl https://api.anthropic.com/v1/sessions/$SESSION_ID/threads/$THREAD_ID/events \2376    -H 'anthropic-version: 2023-06-01' \2377    -H 'anthropic-beta: managed-agents-2026-04-01' \2378    -H "X-Api-Key: $ANTHROPIC_API_KEY"2379```23802381#### Response (200)23822383```json2384{2385  "data": [2386    {2387      "id": "sevt_011CZkZGOp0iBcp4kaQSihUmy",2388      "content": [2389        {2390          "text": "Where is my order #1234?",2391          "type": "text"2392        }2393      ],2394      "type": "user.message",2395      "processed_at": "2026-03-15T10:00:00Z"2396    }2397  ],2398  "next_page": "next_page"2399}2400```24012402## Stream Session Thread Events24032404**GET** `/v1/sessions/{session_id}/threads/{thread_id}/stream`24052406Stream Session Thread Events24072408### Path parameters24092410- `session_id: string`24112412- `thread_id: string`24132414### Query parameters24152416- `event_deltas: optional array of BetaManagedAgentsDeltaType`24172418  When set, this connection also receives streaming deltas (`event_start`, `event_delta`) while an event is being produced, before the event itself arrives. Deltas are best-effort; when the final event is produced it carries the complete content. A model request that ends early (an error or interrupt) produces no final event — its terminal `span.model_request_end` closes the preview. Accepts one or more event types to preview and may be repeated: `agent.message` streams `content_delta` fragments; `agent.thinking` is start-only — a signal that the agent has begun extended thinking, concluded by the `agent.thinking` event itself. Only previews of the requested event types are sent.24192420  - `"agent.message"`24212422  - `"agent.thinking"`24232424### Headers24252426- `"anthropic-beta": optional array of AnthropicBeta`24272428  Optional header to specify the beta version(s) you want to use.24292430  - `string`24312432  - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 43 more`24332434    - `"message-batches-2024-09-24"`24352436    - `"prompt-caching-2024-07-31"`24372438    - `"computer-use-2024-10-22"`24392440    - `"computer-use-2025-01-24"`24412442    - `"pdfs-2024-09-25"`24432444    - `"token-counting-2024-11-01"`24452446    - `"token-efficient-tools-2025-02-19"`24472448    - `"output-128k-2025-02-19"`24492450    - `"files-api-2025-04-14"`24512452    - `"mcp-client-2025-04-04"`24532454    - `"mcp-client-2025-11-20"`24552456    - `"dev-full-thinking-2025-05-14"`24572458    - `"interleaved-thinking-2025-05-14"`24592460    - `"code-execution-2025-05-22"`24612462    - `"extended-cache-ttl-2025-04-11"`24632464    - `"context-1m-2025-08-07"`24652466    - `"context-management-2025-06-27"`24672468    - `"model-context-window-exceeded-2025-08-26"`24692470    - `"skills-2025-10-02"`24712472    - `"fast-mode-2026-02-01"`24732474    - `"output-300k-2026-03-24"`24752476    - `"user-profiles-2026-03-24"`24772478    - `"user-profiles-2026-08-18"`24792480    - `"user-profiles-2026-09-04"`24812482    - `"advisor-tool-2026-03-01"`24832484    - `"managed-agents-2026-04-01"`24852486    - `"cache-diagnosis-2026-04-07"`24872488    - `"dreaming-2026-04-21"`24892490    - `"thinking-token-count-2026-05-13"`24912492    - `"server-side-fallback-2026-06-01"`24932494    - `"server-side-fallback-2026-07-01"`24952496    - `"fallback-credit-2026-06-01"`24972498    - `"fallback-credit-2026-07-01"`24992500    - `"agent-memory-2026-07-22"`25012502    - `"mid-conversation-tool-changes-2026-07-01"`25032504    - `"compact-2026-01-12"`25052506    - `"computer-use-2025-11-24"`25072508    - `"mcp-tunnels-2026-06-22"`25092510    - `"structured-outputs-2025-11-13"`25112512    - `"task-budgets-2026-03-13"`25132514    - `"thinking-display-updates-2026-08-18"`25152516    - `"ce-user-management-2026-07-13"`25172518    - `"mid-conversation-output-config-2026-07-01"`25192520    - `"thinking-binding-controls-2026-08-01"`25212522    - `"mid-conversation-system-clear-at-2026-08-21"`25232524    - `"compact-2026-09-04"`25252526- `"anthropic-workspace-id": optional string`25272528### Returns25292530- `BetaManagedAgentsStreamSessionThreadEvents = BetaManagedAgentsUserMessageEvent or BetaManagedAgentsUserInterruptEvent or BetaManagedAgentsUserToolConfirmationEvent or 34 more`25312532  Server-sent event in a single thread's stream.25332534  - `BetaManagedAgentsUserMessageEvent object`25352536    A user message event in the session conversation.25372538    - `type: "user.message"`25392540    - `id: string`25412542      Unique identifier for this event.25432544    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`25452546      Array of content blocks comprising the user message.25472548      - `BetaManagedAgentsTextBlock object`25492550        Regular text content.25512552        - `type: "text"`25532554        - `text: string`25552556          The text content.25572558          minLength: 125592560      - `BetaManagedAgentsImageBlock object`25612562        Image content specified directly as base64 data or as a reference via a URL.25632564        - `type: "image"`25652566        - `source: BetaManagedAgentsBase64ImageSource or BetaManagedAgentsURLImageSource or BetaManagedAgentsFileImageSource`25672568          Union type for image source variants.25692570          - `BetaManagedAgentsBase64ImageSource object`25712572            Base64-encoded image data.25732574            - `type: "base64"`25752576            - `data: string`25772578              Base64-encoded image data.25792580              minLength: 125812582            - `media_type: string`25832584              MIME type of the image (e.g., "image/png", "image/jpeg", "image/gif", "image/webp").25852586              minLength: 125872588          - `BetaManagedAgentsURLImageSource object`25892590            Image referenced by URL.25912592            - `type: "url"`25932594            - `url: string`25952596              URL of the image to fetch.25972598              minLength: 125992600          - `BetaManagedAgentsFileImageSource object`26012602            Image referenced by file ID.26032604            - `type: "file"`26052606            - `file_id: string`26072608              ID of a previously uploaded file.26092610              minLength: 126112612      - `BetaManagedAgentsDocumentBlock object`26132614        Document content, either specified directly as base64 data, as text, or as a reference via a URL.26152616        - `type: "document"`26172618        - `source: BetaManagedAgentsBase64DocumentSource or BetaManagedAgentsPlainTextDocumentSource or BetaManagedAgentsURLDocumentSource or BetaManagedAgentsFileDocumentSource`26192620          Union type for document source variants.26212622          - `BetaManagedAgentsBase64DocumentSource object`26232624            Base64-encoded document data.26252626            - `type: "base64"`26272628            - `data: string`26292630              Base64-encoded document data.26312632              minLength: 126332634            - `media_type: string`26352636              MIME type of the document (e.g., "application/pdf").26372638              minLength: 126392640          - `BetaManagedAgentsPlainTextDocumentSource object`26412642            Plain text document content.26432644            - `type: "text"`26452646            - `data: string`26472648              The plain text content.26492650              minLength: 126512652            - `media_type: "text/plain"`26532654              MIME type of the text content. Must be "text/plain".26552656          - `BetaManagedAgentsURLDocumentSource object`26572658            Document referenced by URL.26592660            - `type: "url"`26612662            - `url: string`26632664              URL of the document to fetch.26652666              minLength: 126672668          - `BetaManagedAgentsFileDocumentSource object`26692670            Document referenced by file ID.26712672            - `type: "file"`26732674            - `file_id: string`26752676              ID of a previously uploaded file.26772678              minLength: 126792680        - `context: optional string or null`26812682          Additional context about the document for the model.26832684        - `title: optional string or null`26852686          The title of the document.26872688      - `BetaManagedAgentsRedactedBlock object`26892690        Placeholder for content withheld by Anthropic model policy.26912692        - `type: "redacted"`26932694    - `processed_at: optional string or null`26952696      A timestamp in RFC 3339 format26972698      format: date-time26992700  - `BetaManagedAgentsUserInterruptEvent object`27012702    An interrupt event that pauses agent execution and returns control to the user.27032704    - `type: "user.interrupt"`27052706    - `id: string`27072708      Unique identifier for this event.27092710    - `processed_at: optional string or null`27112712      A timestamp in RFC 3339 format27132714      format: date-time27152716    - `session_thread_id: optional string or null`27172718      If absent, interrupts every non-archived thread in a multiagent session (or the primary alone in a single-agent session). If present, interrupts only the named thread.27192720  - `BetaManagedAgentsUserToolConfirmationEvent object`27212722    A tool confirmation event that approves or denies a pending tool execution.27232724    - `type: "user.tool_confirmation"`27252726    - `id: string`27272728      Unique identifier for this event.27292730    - `result: "allow" or "deny"`27312732      UserToolConfirmationResult enum27332734      - `"allow"`27352736      - `"deny"`27372738    - `tool_use_id: string`27392740      The id of the `agent.tool_use` or `agent.mcp_tool_use` event this result corresponds to, which can be found in the last `session.status_idle` [event's](https://platform.claude.com/docs/en/api/beta/sessions/events/list#beta_managed_agents_session_requires_action.event_ids) `stop_reason.event_ids` field.27412742    - `deny_message: optional string or null`27432744      Optional message providing context for a 'deny' decision. Only allowed when result is 'deny'.27452746      maxLength: 1000027472748    - `processed_at: optional string or null`27492750      A timestamp in RFC 3339 format27512752      format: date-time27532754    - `session_thread_id: optional string or null`27552756      Set by the server to the subagent thread this confirmation was routed to. Omitted when it was routed to the primary thread.27572758  - `BetaManagedAgentsUserCustomToolResultEvent object`27592760    Event sent by the client providing the result of a custom tool execution.27612762    - `type: "user.custom_tool_result"`27632764    - `id: string`27652766      Unique identifier for this event.27672768    - `custom_tool_use_id: string`27692770      The id of the `agent.custom_tool_use` event this result corresponds to, which can be found in the last `session.status_idle` [event's](https://platform.claude.com/docs/en/api/beta/sessions/events/list#beta_managed_agents_session_requires_action.event_ids) `stop_reason.event_ids` field.27712772    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`27732774      The result content returned by the tool.27752776      - `BetaManagedAgentsTextBlock object`27772778        Regular text content.27792780      - `BetaManagedAgentsImageBlock object`27812782        Image content specified directly as base64 data or as a reference via a URL.27832784      - `BetaManagedAgentsDocumentBlock object`27852786        Document content, either specified directly as base64 data, as text, or as a reference via a URL.27872788      - `BetaManagedAgentsSearchResultBlock object`27892790        A block containing a web search result.27912792        - `type: "search_result"`27932794        - `citations: BetaManagedAgentsSearchResultCitations`27952796          Citation settings for a search result.27972798          - `enabled: boolean`27992800            Whether citations are enabled for this search result.28012802        - `content: array of BetaManagedAgentsSearchResultContent`28032804          Array of text content blocks from the search result.28052806          - `type: "text"`28072808          - `text: string`28092810            The text content.28112812            minLength: 128132814        - `source: string`28152816          The URL source of the search result.28172818          minLength: 128192820        - `title: string`28212822          The title of the search result.28232824          minLength: 128252826    - `is_error: optional boolean or null`28272828      Whether the tool execution resulted in an error.28292830    - `processed_at: optional string or null`28312832      A timestamp in RFC 3339 format28332834      format: date-time28352836    - `session_thread_id: optional string or null`28372838      Set by the server to the subagent thread this result was routed to. Omitted when it was routed to the primary thread.28392840  - `BetaManagedAgentsAgentCustomToolUseEvent object`28412842    Event emitted when the agent calls a custom tool. The session goes idle until the client sends a `user.custom_tool_result` event with the result.28432844    - `type: "agent.custom_tool_use"`28452846    - `id: string`28472848      Unique identifier for this event.28492850    - `input: map[unknown]`28512852      Input parameters for the tool call.28532854    - `name: string`28552856      Name of the custom tool being called.28572858    - `processed_at: string`28592860      A timestamp in RFC 3339 format28612862      format: date-time28632864    - `session_thread_id: optional string or null`28652866      When set, this event was cross-posted from a subagent's thread to surface its custom tool use on the primary thread's stream. Empty on the thread's own events. Informational only: the server routes the matching `user.custom_tool_result` by `custom_tool_use_id`, so clients do not send it back.28672868  - `BetaManagedAgentsAgentMessageEvent object`28692870    An agent response event in the session conversation.28712872    - `type: "agent.message"`28732874    - `id: string`28752876      Unique identifier for this event.28772878    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsRedactedBlock`28792880      Array of text blocks comprising the agent response.28812882      - `BetaManagedAgentsTextBlock object`28832884        Regular text content.28852886      - `BetaManagedAgentsRedactedBlock object`28872888        Placeholder for content withheld by Anthropic model policy.28892890    - `processed_at: string`28912892      A timestamp in RFC 3339 format28932894      format: date-time28952896  - `BetaManagedAgentsAgentThinkingEvent object`28972898    Indicates the agent is making forward progress via extended thinking. A progress signal, not a content carrier.28992900    - `type: "agent.thinking"`29012902    - `id: string`29032904      Unique identifier for this event.29052906    - `processed_at: string`29072908      A timestamp in RFC 3339 format29092910      format: date-time29112912  - `BetaManagedAgentsAgentMCPToolUseEvent object`29132914    Event emitted when the agent invokes a tool provided by an MCP server.29152916    - `type: "agent.mcp_tool_use"`29172918    - `id: string`29192920      Unique identifier for this event.29212922    - `input: map[unknown]`29232924      Input parameters for the tool call.29252926    - `mcp_server_name: string`29272928      Name of the MCP server providing the tool.29292930    - `name: string`29312932      Name of the MCP tool being used.29332934    - `processed_at: string`29352936      A timestamp in RFC 3339 format29372938      format: date-time29392940    - `evaluated_permission: optional "allow" or "ask" or "deny"`29412942      AgentEvaluatedPermission enum29432944      - `"allow"`29452946      - `"ask"`29472948      - `"deny"`29492950    - `evaluation: optional BetaManagedAgentsAgentToolEvaluation`29512952      Names the resolved permission_policy that produced evaluated_permission, and under auto carries the judgement. Open union: clients must tolerate unknown variants.29532954      - `BetaManagedAgentsAgentToolEvaluationAlwaysAllow object`29552956        The resolved permission_policy was always_allow; accompanies evaluated_permission "allow".29572958        - `type: "always_allow"`29592960      - `BetaManagedAgentsAgentToolEvaluationAlwaysAsk object`29612962        The resolved permission_policy was always_ask; accompanies evaluated_permission "ask".29632964        - `type: "always_ask"`29652966      - `BetaManagedAgentsAgentToolEvaluationAuto object`29672968        The resolved permission_policy was auto: the server judged this invocation individually.29692970        - `type: "auto"`29712972        - `evaluated_permission: BetaManagedAgentsAgentAutoEvaluatedPermission`29732974          The server's per-invocation judgement under the auto permission policy. Its type always equals the event's top-level evaluated_permission. Open union: clients must tolerate unknown variants.29752976          - `BetaManagedAgentsAgentAutoEvaluatedPermissionAllow object`29772978            The server judged the invocation safe to execute without client approval.29792980            - `type: "allow"`29812982          - `BetaManagedAgentsAgentAutoEvaluatedPermissionAsk object`29832984            The server reached no judgement; the invocation is held for client approval.29852986            - `type: "ask"`29872988            - `reason_code: string`29892990              The judgement's grounds in registry-bound terms, for client branching and audit rather than end-user display. Open registry; currently "indeterminate" (no judgement was reached). Clients must tolerate values outside this set.29912992              maxLength: 6429932994          - `BetaManagedAgentsAgentAutoEvaluatedPermissionDeny object`29952996            The server judged the invocation high-risk; it does not execute and a synthetic error tool result is appended.29972998            - `type: "deny"`29993000            - `reason_code: string`30013002              The judgement's grounds in registry-bound terms. Open registry; currently "high_risk" (judged high-risk; the call does not run). Clients must tolerate values outside this set.30033004              maxLength: 6430053006    - `session_thread_id: optional string or null`30073008      When set, this event was cross-posted from a subagent's thread to surface its permission request on the primary thread's stream. Empty on the thread's own events. Informational only: the server routes the matching `user.tool_confirmation` by `tool_use_id`, so clients do not send it back.30093010  - `BetaManagedAgentsAgentMCPToolResultEvent object`30113012    Event representing the result of an MCP tool execution.30133014    - `type: "agent.mcp_tool_result"`30153016    - `id: string`30173018      Unique identifier for this event.30193020    - `mcp_tool_use_id: string`30213022      The id of the `agent.mcp_tool_use` event this result corresponds to.30233024    - `processed_at: string`30253026      A timestamp in RFC 3339 format30273028      format: date-time30293030    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`30313032      The result content returned by the tool.30333034      - `BetaManagedAgentsTextBlock object`30353036        Regular text content.30373038      - `BetaManagedAgentsImageBlock object`30393040        Image content specified directly as base64 data or as a reference via a URL.30413042      - `BetaManagedAgentsDocumentBlock object`30433044        Document content, either specified directly as base64 data, as text, or as a reference via a URL.30453046      - `BetaManagedAgentsSearchResultBlock object`30473048        A block containing a web search result.30493050    - `is_error: optional boolean or null`30513052      Whether the tool execution resulted in an error.30533054  - `BetaManagedAgentsAgentToolUseEvent object`30553056    Event emitted when the agent invokes a built-in agent tool.30573058    - `type: "agent.tool_use"`30593060    - `id: string`30613062      Unique identifier for this event.30633064    - `input: map[unknown]`30653066      Input parameters for the tool call.30673068    - `name: string`30693070      Name of the agent tool being used.30713072    - `processed_at: string`30733074      A timestamp in RFC 3339 format30753076      format: date-time30773078    - `evaluated_permission: optional "allow" or "ask" or "deny"`30793080      AgentEvaluatedPermission enum30813082      - `"allow"`30833084      - `"ask"`30853086      - `"deny"`30873088    - `evaluation: optional BetaManagedAgentsAgentToolEvaluation`30893090      Names the resolved permission_policy that produced evaluated_permission, and under auto carries the judgement. Open union: clients must tolerate unknown variants.30913092    - `session_thread_id: optional string or null`30933094      When set, this event was cross-posted from a subagent's thread to surface its permission request on the primary thread's stream. Empty on the thread's own events. Informational only: the server routes the matching `user.tool_confirmation` or `user.tool_result` by `tool_use_id`, so clients do not send it back.30953096  - `BetaManagedAgentsAgentToolResultEvent object`30973098    Event representing the result of an agent tool execution.30993100    - `type: "agent.tool_result"`31013102    - `id: string`31033104      Unique identifier for this event.31053106    - `processed_at: string`31073108      A timestamp in RFC 3339 format31093110      format: date-time31113112    - `tool_use_id: string`31133114      The id of the `agent.tool_use` event this result corresponds to.31153116    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`31173118      The result content returned by the tool.31193120      - `BetaManagedAgentsTextBlock object`31213122        Regular text content.31233124      - `BetaManagedAgentsImageBlock object`31253126        Image content specified directly as base64 data or as a reference via a URL.31273128      - `BetaManagedAgentsDocumentBlock object`31293130        Document content, either specified directly as base64 data, as text, or as a reference via a URL.31313132      - `BetaManagedAgentsSearchResultBlock object`31333134        A block containing a web search result.31353136    - `is_error: optional boolean or null`31373138      Whether the tool execution resulted in an error.31393140  - `BetaManagedAgentsAgentThreadMessageReceivedEvent object`31413142    Delivery event written to the target thread's input stream when an agent-to-agent message arrives.31433144    - `type: "agent.thread_message_received"`31453146    - `id: string`31473148      Unique identifier for this event.31493150    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`31513152      Message content blocks.31533154      - `BetaManagedAgentsTextBlock object`31553156        Regular text content.31573158      - `BetaManagedAgentsImageBlock object`31593160        Image content specified directly as base64 data or as a reference via a URL.31613162      - `BetaManagedAgentsDocumentBlock object`31633164        Document content, either specified directly as base64 data, as text, or as a reference via a URL.31653166      - `BetaManagedAgentsRedactedBlock object`31673168        Placeholder for content withheld by Anthropic model policy.31693170    - `from_session_thread_id: string`31713172      Public `sthr_` ID of the thread that sent the message.31733174    - `processed_at: string`31753176      A timestamp in RFC 3339 format31773178      format: date-time31793180    - `from_agent_name: optional string or null`31813182      Name of the callable agent this message came from. Absent when received from the primary agent.31833184  - `BetaManagedAgentsAgentThreadMessageSentEvent object`31853186    Observability event emitted to the sender's output stream when an agent-to-agent message is sent.31873188    - `type: "agent.thread_message_sent"`31893190    - `id: string`31913192      Unique identifier for this event.31933194    - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`31953196      Message content blocks.31973198      - `BetaManagedAgentsTextBlock object`31993200        Regular text content.32013202      - `BetaManagedAgentsImageBlock object`32033204        Image content specified directly as base64 data or as a reference via a URL.32053206      - `BetaManagedAgentsDocumentBlock object`32073208        Document content, either specified directly as base64 data, as text, or as a reference via a URL.32093210      - `BetaManagedAgentsRedactedBlock object`32113212        Placeholder for content withheld by Anthropic model policy.32133214    - `processed_at: string`32153216      A timestamp in RFC 3339 format32173218      format: date-time32193220    - `to_session_thread_id: string`32213222      Public `sthr_` ID of the thread the message was sent to.32233224    - `to_agent_name: optional string or null`32253226      Name of the callable agent this message was sent to. Absent when sent to the primary agent.32273228  - `BetaManagedAgentsAgentThreadContextCompactedEvent object`32293230    Indicates that context compaction (summarization) occurred during the session.32313232    - `type: "agent.thread_context_compacted"`32333234    - `id: string`32353236      Unique identifier for this event.32373238    - `processed_at: string`32393240      A timestamp in RFC 3339 format32413242      format: date-time32433244  - `BetaManagedAgentsSessionErrorEvent object`32453246    An error event indicating a problem occurred during session execution.32473248    - `type: "session.error"`32493250    - `id: string`32513252      Unique identifier for this event.32533254    - `error: BetaManagedAgentsUnknownError or BetaManagedAgentsModelOverloadedError or BetaManagedAgentsModelRateLimitedError or 5 more`32553256      - `BetaManagedAgentsUnknownError object`32573258        An unknown or unexpected error occurred during session execution. A fallback variant; clients that don't recognize a new error code can match on `retry_status` and `message` alone.32593260        - `type: "unknown_error"`32613262        - `message: string`32633264          Human-readable error description.32653266        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`32673268          What the client should do next in response to this error.32693270          - `BetaManagedAgentsRetryStatusRetrying object`32713272            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.32733274            - `type: "retrying"`32753276          - `BetaManagedAgentsRetryStatusExhausted object`32773278            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.32793280            - `type: "exhausted"`32813282          - `BetaManagedAgentsRetryStatusTerminal object`32833284            The session encountered a terminal error and will transition to `terminated` state.32853286            - `type: "terminal"`32873288      - `BetaManagedAgentsModelOverloadedError object`32893290        The model is currently overloaded. Emitted after automatic retries are exhausted.32913292        - `type: "model_overloaded_error"`32933294        - `message: string`32953296          Human-readable error description.32973298        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`32993300          What the client should do next in response to this error.33013302          - `BetaManagedAgentsRetryStatusRetrying object`33033304            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.33053306          - `BetaManagedAgentsRetryStatusExhausted object`33073308            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.33093310          - `BetaManagedAgentsRetryStatusTerminal object`33113312            The session encountered a terminal error and will transition to `terminated` state.33133314      - `BetaManagedAgentsModelRateLimitedError object`33153316        The model request was rate-limited.33173318        - `type: "model_rate_limited_error"`33193320        - `message: string`33213322          Human-readable error description.33233324        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`33253326          What the client should do next in response to this error.33273328          - `BetaManagedAgentsRetryStatusRetrying object`33293330            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.33313332          - `BetaManagedAgentsRetryStatusExhausted object`33333334            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.33353336          - `BetaManagedAgentsRetryStatusTerminal object`33373338            The session encountered a terminal error and will transition to `terminated` state.33393340      - `BetaManagedAgentsModelRequestFailedError object`33413342        A model request failed for a reason other than overload or rate-limiting.33433344        - `type: "model_request_failed_error"`33453346        - `message: string`33473348          Human-readable error description.33493350        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`33513352          What the client should do next in response to this error.33533354          - `BetaManagedAgentsRetryStatusRetrying object`33553356            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.33573358          - `BetaManagedAgentsRetryStatusExhausted object`33593360            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.33613362          - `BetaManagedAgentsRetryStatusTerminal object`33633364            The session encountered a terminal error and will transition to `terminated` state.33653366      - `BetaManagedAgentsMCPConnectionFailedError object`33673368        Failed to connect to an MCP server.33693370        - `type: "mcp_connection_failed_error"`33713372        - `mcp_server_name: string`33733374          Name of the MCP server that failed to connect.33753376        - `message: string`33773378          Human-readable error description.33793380        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`33813382          What the client should do next in response to this error.33833384          - `BetaManagedAgentsRetryStatusRetrying object`33853386            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.33873388          - `BetaManagedAgentsRetryStatusExhausted object`33893390            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.33913392          - `BetaManagedAgentsRetryStatusTerminal object`33933394            The session encountered a terminal error and will transition to `terminated` state.33953396      - `BetaManagedAgentsMCPAuthenticationFailedError object`33973398        Authentication to an MCP server failed.33993400        - `type: "mcp_authentication_failed_error"`34013402        - `mcp_server_name: string`34033404          Name of the MCP server that failed authentication.34053406        - `message: string`34073408          Human-readable error description.34093410        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`34113412          What the client should do next in response to this error.34133414          - `BetaManagedAgentsRetryStatusRetrying object`34153416            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.34173418          - `BetaManagedAgentsRetryStatusExhausted object`34193420            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.34213422          - `BetaManagedAgentsRetryStatusTerminal object`34233424            The session encountered a terminal error and will transition to `terminated` state.34253426      - `BetaManagedAgentsBillingError object`34273428        The caller's organization or workspace cannot make model requests — out of credits or spend limit reached. Retrying with the same credentials will not succeed; the caller must resolve the billing state.34293430        - `type: "billing_error"`34313432        - `message: string`34333434          Human-readable error description.34353436        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`34373438          What the client should do next in response to this error.34393440          - `BetaManagedAgentsRetryStatusRetrying object`34413442            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.34433444          - `BetaManagedAgentsRetryStatusExhausted object`34453446            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.34473448          - `BetaManagedAgentsRetryStatusTerminal object`34493450            The session encountered a terminal error and will transition to `terminated` state.34513452      - `BetaManagedAgentsCredentialHostUnreachableError object`34533454        An `environment_variable` credential's `auth.networking.allowed_hosts` includes a host the environment's network policy does not permit.34553456        - `type: "credential_host_unreachable_error"`34573458        - `credential_id: string`34593460          ID of the affected credential.34613462        - `message: string`34633464          Human-readable error description.34653466        - `retry_status: BetaManagedAgentsRetryStatusRetrying or BetaManagedAgentsRetryStatusExhausted or BetaManagedAgentsRetryStatusTerminal`34673468          What the client should do next in response to this error.34693470          - `BetaManagedAgentsRetryStatusRetrying object`34713472            The server is retrying automatically. Client should wait; the same error type may fire again as retrying, then once as exhausted when the retry budget runs out.34733474          - `BetaManagedAgentsRetryStatusExhausted object`34753476            This turn is dead; queued inputs are flushed and the session returns to idle. Client may send a new prompt.34773478          - `BetaManagedAgentsRetryStatusTerminal object`34793480            The session encountered a terminal error and will transition to `terminated` state.34813482        - `vault_id: string`34833484          ID of the vault containing the affected credential.34853486    - `processed_at: string`34873488      A timestamp in RFC 3339 format34893490      format: date-time34913492  - `BetaManagedAgentsSessionStatusRescheduledEvent object`34933494    Indicates the session is recovering from an error state and is rescheduled for execution.34953496    - `type: "session.status_rescheduled"`34973498    - `id: string`34993500      Unique identifier for this event.35013502    - `processed_at: string`35033504      A timestamp in RFC 3339 format35053506      format: date-time35073508  - `BetaManagedAgentsSessionStatusRunningEvent object`35093510    Indicates the session is actively running and the agent is working.35113512    - `type: "session.status_running"`35133514    - `id: string`35153516      Unique identifier for this event.35173518    - `processed_at: string`35193520      A timestamp in RFC 3339 format35213522      format: date-time35233524  - `BetaManagedAgentsSessionStatusIdleEvent object`35253526    Indicates the agent has paused and is awaiting user input.35273528    - `type: "session.status_idle"`35293530    - `id: string`35313532      Unique identifier for this event.35333534    - `processed_at: string`35353536      A timestamp in RFC 3339 format35373538      format: date-time35393540    - `stop_reason: BetaManagedAgentsSessionEndTurn or BetaManagedAgentsSessionRequiresAction or BetaManagedAgentsSessionRetriesExhausted or BetaManagedAgentsSessionBudgetReached`35413542      - `BetaManagedAgentsSessionEndTurn object`35433544        The agent completed its turn naturally and is ready for the next user message.35453546        - `type: "end_turn"`35473548      - `BetaManagedAgentsSessionRequiresAction object`35493550        The agent is idle waiting on one or more blocking user-input events (tool confirmation, custom tool result, etc.). Resolving all of them transitions the session back to running.35513552        - `type: "requires_action"`35533554        - `event_ids: array of string`35553556          The ids of events the agent is blocked on. Resolving fewer than all re-emits `session.status_idle` with the remainder.35573558      - `BetaManagedAgentsSessionRetriesExhausted object`35593560        The turn ended because repeated errors exhausted the retry budget or an error escalated to `retry_status: 'exhausted'`.35613562        - `type: "retries_exhausted"`35633564      - `BetaManagedAgentsSessionBudgetReached object`35653566        The agent stopped because the session's tracked list cost reached its budget, or because its usage includes a model with no list price (which the budget cannot measure). Raise the budget to continue — or, if raising is rejected because a model has no list price, remove the budget.35673568        - `type: "budget_reached"`35693570  - `BetaManagedAgentsSessionStatusTerminatedEvent object`35713572    Indicates the session has terminated, either due to an error or completion.35733574    - `type: "session.status_terminated"`35753576    - `id: string`35773578      Unique identifier for this event.35793580    - `processed_at: string`35813582      A timestamp in RFC 3339 format35833584      format: date-time35853586  - `BetaManagedAgentsSessionThreadCreatedEvent object`35873588    Emitted when a subagent is spawned as a new thread. Written to the parent thread's output stream so clients observing the session see child creation.35893590    - `type: "session.thread_created"`35913592    - `id: string`35933594      Unique identifier for this event.35953596    - `agent_name: string`35973598      Name of the callable agent the thread runs.35993600    - `processed_at: string`36013602      A timestamp in RFC 3339 format36033604      format: date-time36053606    - `session_thread_id: string`36073608      Public `sthr_` ID of the newly created thread.36093610  - `BetaManagedAgentsSpanOutcomeEvaluationStartEvent object`36113612    Emitted when an outcome evaluation cycle begins.36133614    - `type: "span.outcome_evaluation_start"`36153616    - `id: string`36173618      Unique identifier for this event.36193620    - `iteration: number`36213622      0-indexed revision cycle. 0 is the first evaluation; 1 is the re-evaluation after the first revision; etc.36233624      format: int3236253626    - `outcome_id: string`36273628      The `outc_` ID of the outcome being evaluated.36293630    - `processed_at: string`36313632      A timestamp in RFC 3339 format36333634      format: date-time36353636  - `BetaManagedAgentsSpanOutcomeEvaluationEndEvent object`36373638    Emitted when an outcome evaluation cycle completes. Carries the verdict and aggregate token usage. A verdict of `needs_revision` means another evaluation cycle follows; `satisfied`, `max_iterations_reached`, `failed`, or `interrupted` are terminal — no further evaluation cycles follow.36393640    - `type: "span.outcome_evaluation_end"`36413642    - `id: string`36433644      Unique identifier for this event.36453646    - `explanation: string`36473648      Human-readable explanation of the verdict. For `needs_revision`, describes which criteria failed and why.36493650    - `iteration: number`36513652      0-indexed revision cycle, matching the corresponding `span.outcome_evaluation_start`.36533654      format: int3236553656    - `outcome_evaluation_start_id: string`36573658      The id of the corresponding `span.outcome_evaluation_start` event.36593660    - `outcome_id: string`36613662      The `outc_` ID of the outcome being evaluated.36633664    - `processed_at: string`36653666      A timestamp in RFC 3339 format36673668      format: date-time36693670    - `result: string`36713672      Evaluation verdict. 'satisfied': criteria met, session goes idle. 'needs_revision': criteria not met, another revision cycle follows. 'max_iterations_reached': evaluation budget exhausted with criteria still unmet — one final acknowledgment turn follows before the session goes idle, but no further evaluation runs. 'failed': grader determined the rubric does not apply to the deliverables. 'interrupted': user sent an interrupt while evaluation was in progress.36733674    - `usage: BetaManagedAgentsSpanModelUsage`36753676      Token usage for a single model request.36773678      - `cache_creation_input_tokens: number`36793680        Tokens used to create prompt cache in this request.36813682        format: int3236833684      - `cache_read_input_tokens: number`36853686        Tokens read from prompt cache in this request.36873688        format: int3236893690      - `input_tokens: number`36913692        Input tokens consumed by this request.36933694        format: int3236953696      - `output_tokens: number`36973698        Output tokens generated by this request.36993700        format: int3237013702      - `speed: optional "standard" or "fast" or null`37033704        Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Not all models support `fast`; invalid combinations are rejected at create time.37053706        - `"standard"`37073708        - `"fast"`37093710  - `BetaManagedAgentsSpanModelRequestStartEvent object`37113712    Emitted when a model request is initiated by the agent.37133714    - `type: "span.model_request_start"`37153716    - `id: string`37173718      Unique identifier for this event.37193720    - `processed_at: string`37213722      A timestamp in RFC 3339 format37233724      format: date-time37253726  - `BetaManagedAgentsSpanModelRequestEndEvent object`37273728    Emitted when a model request completes.37293730    - `type: "span.model_request_end"`37313732    - `id: string`37333734      Unique identifier for this event.37353736    - `is_error: boolean or null`37373738      Whether the model request resulted in an error.37393740    - `model_request_start_id: string`37413742      The id of the corresponding `span.model_request_start` event.37433744    - `model_usage: BetaManagedAgentsSpanModelUsage`37453746      Token usage for a single model request.37473748    - `processed_at: string`37493750      A timestamp in RFC 3339 format37513752      format: date-time37533754  - `BetaManagedAgentsSpanOutcomeEvaluationOngoingEvent object`37553756    Periodic heartbeat emitted while an outcome evaluation cycle is in progress. Distinguishes 'evaluation is actively running' from 'evaluation is stuck' between the corresponding `span.outcome_evaluation_start` and `span.outcome_evaluation_end` events.37573758    - `type: "span.outcome_evaluation_ongoing"`37593760    - `id: string`37613762      Unique identifier for this event.37633764    - `iteration: number`37653766      0-indexed revision cycle, matching the corresponding `span.outcome_evaluation_start`.37673768      format: int3237693770    - `outcome_id: string`37713772      The `outc_` ID of the outcome being evaluated.37733774    - `processed_at: string`37753776      A timestamp in RFC 3339 format37773778      format: date-time37793780  - `BetaManagedAgentsUserDefineOutcomeEvent object`37813782    Echo of a `user.define_outcome` input event. Carries the server-generated `outcome_id` that subsequent `span.outcome_evaluation_*` events reference.37833784    - `type: "user.define_outcome"`37853786    - `id: string`37873788      Unique identifier for this event.37893790    - `description: string`37913792      What the agent should produce. Copied from the input event.37933794    - `max_iterations: number or null`37953796      Evaluate-then-revise cycles before giving up. Default 3, max 20.37973798      format: int3237993800    - `outcome_id: string`38013802      Server-generated `outc_` ID for this outcome. Referenced by `span.outcome_evaluation_*` events and the session's `outcome_evaluations` list.38033804    - `processed_at: string`38053806      A timestamp in RFC 3339 format38073808      format: date-time38093810    - `rubric: BetaManagedAgentsFileRubric or BetaManagedAgentsTextRubric`38113812      Rubric for grading the quality of an outcome.38133814      - `BetaManagedAgentsFileRubric object`38153816        Rubric referenced by a file uploaded via the Files API.38173818        - `type: "file"`38193820        - `file_id: string`38213822          ID of the rubric file.38233824      - `BetaManagedAgentsTextRubric object`38253826        Rubric content provided inline as text.38273828        - `type: "text"`38293830        - `content: string`38313832          Rubric content. Plain text or markdown — the grader treats it as freeform text.38333834  - `BetaManagedAgentsSessionDeletedEvent object`38353836    Emitted when a session has been deleted. Terminates any active event stream — no further events will be emitted for this session.38373838    - `type: "session.deleted"`38393840    - `id: string`38413842      Unique identifier for this event.38433844    - `processed_at: string`38453846      A timestamp in RFC 3339 format38473848      format: date-time38493850  - `BetaManagedAgentsSessionThreadStatusRunningEvent object`38513852    A session thread has begun executing. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.38533854    - `type: "session.thread_status_running"`38553856    - `id: string`38573858      Unique identifier for this event.38593860    - `agent_name: string`38613862      Name of the agent the thread runs.38633864    - `processed_at: string`38653866      A timestamp in RFC 3339 format38673868      format: date-time38693870    - `session_thread_id: string`38713872      Public sthr_ ID of the thread that started running.38733874  - `BetaManagedAgentsSessionThreadStatusIdleEvent object`38753876    A session thread has yielded and is awaiting input. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.38773878    - `type: "session.thread_status_idle"`38793880    - `id: string`38813882      Unique identifier for this event.38833884    - `agent_name: string`38853886      Name of the agent the thread runs.38873888    - `processed_at: string`38893890      A timestamp in RFC 3339 format38913892      format: date-time38933894    - `session_thread_id: string`38953896      Public sthr_ ID of the thread that went idle.38973898    - `stop_reason: BetaManagedAgentsSessionEndTurn or BetaManagedAgentsSessionRequiresAction or BetaManagedAgentsSessionRetriesExhausted or BetaManagedAgentsSessionBudgetReached`38993900      - `BetaManagedAgentsSessionEndTurn object`39013902        The agent completed its turn naturally and is ready for the next user message.39033904      - `BetaManagedAgentsSessionRequiresAction object`39053906        The agent is idle waiting on one or more blocking user-input events (tool confirmation, custom tool result, etc.). Resolving all of them transitions the session back to running.39073908      - `BetaManagedAgentsSessionRetriesExhausted object`39093910        The turn ended because repeated errors exhausted the retry budget or an error escalated to `retry_status: 'exhausted'`.39113912      - `BetaManagedAgentsSessionBudgetReached object`39133914        The agent stopped because the session's tracked list cost reached its budget, or because its usage includes a model with no list price (which the budget cannot measure). Raise the budget to continue — or, if raising is rejected because a model has no list price, remove the budget.39153916  - `BetaManagedAgentsSessionThreadStatusTerminatedEvent object`39173918    A session thread has terminated and will accept no further input. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.39193920    - `type: "session.thread_status_terminated"`39213922    - `id: string`39233924      Unique identifier for this event.39253926    - `agent_name: string`39273928      Name of the agent the thread runs.39293930    - `processed_at: string`39313932      A timestamp in RFC 3339 format39333934      format: date-time39353936    - `session_thread_id: string`39373938      Public sthr_ ID of the thread that terminated.39393940  - `BetaManagedAgentsUserToolResultEvent object`39413942    Event sent by the client providing the result of an agent-toolset tool execution. Only valid on `self_hosted` environments, where sandbox-routed tools are executed by the client rather than the server.39433944    - `type: "user.tool_result"`39453946    - `id: string`39473948      Unique identifier for this event.39493950    - `tool_use_id: string`39513952      The id of the `agent.tool_use` event this result corresponds to, which can be found in the last `session.status_idle` [event's](https://platform.claude.com/docs/en/api/beta/sessions/events/list#beta_managed_agents_session_requires_action.event_ids) `stop_reason.event_ids` field.39533954    - `content: optional array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsSearchResultBlock`39553956      The result content returned by the tool.39573958      - `BetaManagedAgentsTextBlock object`39593960        Regular text content.39613962      - `BetaManagedAgentsImageBlock object`39633964        Image content specified directly as base64 data or as a reference via a URL.39653966      - `BetaManagedAgentsDocumentBlock object`39673968        Document content, either specified directly as base64 data, as text, or as a reference via a URL.39693970      - `BetaManagedAgentsSearchResultBlock object`39713972        A block containing a web search result.39733974    - `is_error: optional boolean or null`39753976      Whether the tool execution resulted in an error.39773978    - `processed_at: optional string or null`39793980      A timestamp in RFC 3339 format39813982      format: date-time39833984    - `session_thread_id: optional string or null`39853986      Set by the server to the subagent thread this result was routed to. Omitted when it was routed to the primary thread.39873988  - `BetaManagedAgentsSessionThreadStatusRescheduledEvent object`39893990    A session thread hit a transient error and is retrying automatically. Emitted on the thread's own stream and cross-posted to the primary stream for child threads.39913992    - `type: "session.thread_status_rescheduled"`39933994    - `id: string`39953996      Unique identifier for this event.39973998    - `agent_name: string`39994000      Name of the agent the thread runs.40014002    - `processed_at: string`40034004      A timestamp in RFC 3339 format40054006      format: date-time40074008    - `session_thread_id: string`40094010      Public sthr_ ID of the thread that is retrying.40114012  - `BetaManagedAgentsSessionUpdatedEvent object`40134014    Emitted when an UpdateSession request changed at least one field. Carries only the fields that changed; absent fields were not part of the update. The new configuration applies from the next turn.40154016    - `type: "session.updated"`40174018    - `id: string`40194020      Unique identifier for this event.40214022    - `processed_at: string`40234024      A timestamp in RFC 3339 format40254026      format: date-time40274028    - `agent: optional BetaManagedAgentsSessionAgent or null`40294030      Resolved `agent` definition for a `session`. Snapshot of the `agent` at `session` creation time.40314032      - `type: "agent"`40334034      - `id: string`40354036      - `description: string or null`40374038      - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`40394040        - `type: "url"`40414042        - `name: string`40434044        - `url: string`40454046      - `model: BetaManagedAgentsModelConfig`40474048        Model identifier and configuration.40494050        - `id: BetaManagedAgentsModel`40514052          The model that will power your agent.40534054          See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.40554056          - `"claude-fable-5-1" or "claude-sonnet-5" or "claude-fable-5" or 11 more`40574058            The model that will power your agent.40594060            See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.40614062            - `"claude-fable-5-1"`40634064              Frontier intelligence for ambitious tasks across coding, scientific discovery, and enterprise workflows40654066            - `"claude-sonnet-5"`40674068              High-performance model for coding and agents40694070            - `"claude-fable-5"`40714072              Next generation of intelligence for the hardest knowledge work and coding problems40734074            - `"claude-opus-5"`40754076              Powerful intelligence for long-running agents and coding40774078            - `"claude-opus-4-8"`40794080              Powerful intelligence for long-running agents and coding40814082            - `"claude-opus-4-7"`40834084              Powerful intelligence for long-running agents and coding40854086            - `"claude-opus-4-6"`40874088              Powerful intelligence for long-running agents and coding40894090            - `"claude-sonnet-4-6"`40914092              Best combination of speed and intelligence40934094            - `"claude-haiku-4-5"`40954096              Fastest model with near-frontier intelligence40974098            - `"claude-haiku-4-5-20251001"`40994100              Fastest model with near-frontier intelligence41014102            - `"claude-opus-4-5"`41034104              Powerful intelligence for long-running agents and coding41054106            - `"claude-opus-4-5-20251101"`41074108              Powerful intelligence for long-running agents and coding41094110            - `"claude-sonnet-4-5"`41114112              High-performance model for agents and coding41134114            - `"claude-sonnet-4-5-20250929"`41154116              High-performance model for agents and coding41174118          - `string`41194120        - `effort: optional BetaManagedAgentsEffortLow or BetaManagedAgentsEffortMedium or BetaManagedAgentsEffortHigh or 2 more`41214122          How hard Claude works on each turn. Sets `output_config.effort` on every Messages call the session makes.41234124          - `BetaManagedAgentsEffortLow object`41254126            Low effort. Favors latency over reasoning depth.41274128            - `type: "low"`41294130          - `BetaManagedAgentsEffortMedium object`41314132            Medium effort. Balances latency and reasoning depth.41334134            - `type: "medium"`41354136          - `BetaManagedAgentsEffortHigh object`41374138            High effort. Favors reasoning depth.41394140            - `type: "high"`41414142          - `BetaManagedAgentsEffortXhigh object`41434144            Extra-high effort. Not all models accept this level.41454146            - `type: "xhigh"`41474148          - `BetaManagedAgentsEffortMax object`41494150            Maximum effort. Favors reasoning depth over latency.41514152            - `type: "max"`41534154        - `inference_geo: optional string`41554156          Geographic region for model inference. When unset, requests fall through to the workspace's default_inference_geo.41574158        - `speed: optional "standard" or "fast"`41594160          Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Not all models support `fast`; invalid combinations are rejected at create time.41614162          - `"standard"`41634164          - `"fast"`41654166      - `multiagent: BetaManagedAgentsSessionMultiagentCoordinator or null`41674168        Resolved coordinator topology with full agent definitions for each roster member.41694170        - `type: "coordinator"`41714172        - `agents: array of BetaManagedAgentsSessionThreadAgent or BetaManagedAgentsAdvisor`41734174          Full `agent` definitions the coordinator may spawn as session threads.41754176          - `BetaManagedAgentsSessionThreadAgent object`41774178            Resolved `agent` definition for a single `session_thread`. Snapshot of the agent at thread creation time. The multiagent roster is not repeated here; read it from `Session.agent`.41794180            - `type: "agent"`41814182            - `id: string`41834184            - `description: string or null`41854186            - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`41874188              - `type: "url"`41894190              - `name: string`41914192              - `url: string`41934194            - `model: BetaManagedAgentsModelConfig`41954196              Model identifier and configuration.41974198            - `name: string`41994200            - `skills: array of BetaManagedAgentsAnthropicSkill or BetaManagedAgentsCustomSkill`42014202              - `BetaManagedAgentsAnthropicSkill object`42034204                A resolved Anthropic-managed skill.42054206                - `type: "anthropic"`42074208                - `skill_id: string`42094210                - `version: string`42114212              - `BetaManagedAgentsCustomSkill object`42134214                A resolved user-created custom skill.42154216                - `type: "custom"`42174218                - `skill_id: string`42194220                - `version: string`42214222            - `system: string or null`42234224            - `tools: array of BetaManagedAgentsAgentToolset20260401 or BetaManagedAgentsMCPToolset or BetaManagedAgentsCustomTool`42254226              - `BetaManagedAgentsAgentToolset20260401 object`42274228                - `type: "agent_toolset_20260401"`42294230                - `configs: array of BetaManagedAgentsAgentToolConfig`42314232                  - `BetaManagedAgentsBashToolConfig object`42334234                    Configuration for the bash tool.42354236                    - `type: "bash"`42374238                    - `enabled: boolean`42394240                    - `name: "bash"`42414242                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`42434244                      Permission policy for tool execution.42454246                      - `BetaManagedAgentsAlwaysAllowPolicy object`42474248                        Tool calls are automatically approved without user confirmation.42494250                        - `type: "always_allow"`42514252                      - `BetaManagedAgentsAlwaysAskPolicy object`42534254                        Tool calls require user confirmation before execution.42554256                        - `type: "always_ask"`42574258                      - `BetaManagedAgentsAutoPolicy object`42594260                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.42614262                        - `type: "auto"`42634264                  - `BetaManagedAgentsEditToolConfig object`42654266                    Configuration for the edit tool.42674268                    - `type: "edit"`42694270                    - `enabled: boolean`42714272                    - `name: "edit"`42734274                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`42754276                      Permission policy for tool execution.42774278                      - `BetaManagedAgentsAlwaysAllowPolicy object`42794280                        Tool calls are automatically approved without user confirmation.42814282                      - `BetaManagedAgentsAlwaysAskPolicy object`42834284                        Tool calls require user confirmation before execution.42854286                      - `BetaManagedAgentsAutoPolicy object`42874288                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.42894290                  - `BetaManagedAgentsReadToolConfig object`42914292                    Configuration for the read tool.42934294                    - `type: "read"`42954296                    - `enabled: boolean`42974298                    - `name: "read"`42994300                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`43014302                      Permission policy for tool execution.43034304                      - `BetaManagedAgentsAlwaysAllowPolicy object`43054306                        Tool calls are automatically approved without user confirmation.43074308                      - `BetaManagedAgentsAlwaysAskPolicy object`43094310                        Tool calls require user confirmation before execution.43114312                      - `BetaManagedAgentsAutoPolicy object`43134314                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.43154316                  - `BetaManagedAgentsWriteToolConfig object`43174318                    Configuration for the write tool.43194320                    - `type: "write"`43214322                    - `enabled: boolean`43234324                    - `name: "write"`43254326                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`43274328                      Permission policy for tool execution.43294330                      - `BetaManagedAgentsAlwaysAllowPolicy object`43314332                        Tool calls are automatically approved without user confirmation.43334334                      - `BetaManagedAgentsAlwaysAskPolicy object`43354336                        Tool calls require user confirmation before execution.43374338                      - `BetaManagedAgentsAutoPolicy object`43394340                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.43414342                  - `BetaManagedAgentsGlobToolConfig object`43434344                    Configuration for the glob tool.43454346                    - `type: "glob"`43474348                    - `enabled: boolean`43494350                    - `name: "glob"`43514352                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`43534354                      Permission policy for tool execution.43554356                      - `BetaManagedAgentsAlwaysAllowPolicy object`43574358                        Tool calls are automatically approved without user confirmation.43594360                      - `BetaManagedAgentsAlwaysAskPolicy object`43614362                        Tool calls require user confirmation before execution.43634364                      - `BetaManagedAgentsAutoPolicy object`43654366                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.43674368                  - `BetaManagedAgentsGrepToolConfig object`43694370                    Configuration for the grep tool.43714372                    - `type: "grep"`43734374                    - `enabled: boolean`43754376                    - `name: "grep"`43774378                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`43794380                      Permission policy for tool execution.43814382                      - `BetaManagedAgentsAlwaysAllowPolicy object`43834384                        Tool calls are automatically approved without user confirmation.43854386                      - `BetaManagedAgentsAlwaysAskPolicy object`43874388                        Tool calls require user confirmation before execution.43894390                      - `BetaManagedAgentsAutoPolicy object`43914392                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.43934394                  - `BetaManagedAgentsWebFetchToolConfig object`43954396                    Configuration for the web_fetch tool.43974398                    - `type: "web_fetch"`43994400                    - `enabled: boolean`44014402                    - `name: "web_fetch"`44034404                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`44054406                      Permission policy for tool execution.44074408                      - `BetaManagedAgentsAlwaysAllowPolicy object`44094410                        Tool calls are automatically approved without user confirmation.44114412                      - `BetaManagedAgentsAlwaysAskPolicy object`44134414                        Tool calls require user confirmation before execution.44154416                      - `BetaManagedAgentsAutoPolicy object`44174418                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.44194420                    - `allowed_domains: optional array of string`44214422                    - `blocked_domains: optional array of string`44234424                    - `max_content_tokens: optional number or null`44254426                      format: int3244274428                  - `BetaManagedAgentsWebSearchToolConfig object`44294430                    Configuration for the web_search tool.44314432                    - `type: "web_search"`44334434                    - `enabled: boolean`44354436                    - `name: "web_search"`44374438                    - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`44394440                      Permission policy for tool execution.44414442                      - `BetaManagedAgentsAlwaysAllowPolicy object`44434444                        Tool calls are automatically approved without user confirmation.44454446                      - `BetaManagedAgentsAlwaysAskPolicy object`44474448                        Tool calls require user confirmation before execution.44494450                      - `BetaManagedAgentsAutoPolicy object`44514452                        The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.44534454                    - `allowed_domains: optional array of string`44554456                    - `blocked_domains: optional array of string`44574458                    - `user_location: optional BetaManagedAgentsUserLocation or null`44594460                      Approximate user location for search result localization.44614462                      - `type: "approximate"`44634464                        Location precision. Only "approximate" is supported.44654466                      - `city: optional string or null`44674468                        City name.44694470                        minLength: 1, maxLength: 25544714472                      - `country: optional string or null`44734474                        Two-letter ISO 3166-1 country code, uppercase.44754476                      - `region: optional string or null`44774478                        Region or state name.44794480                        minLength: 1, maxLength: 25544814482                      - `timezone: optional string or null`44834484                        IANA timezone identifier, e.g. "America/Los_Angeles".44854486                        minLength: 1, maxLength: 25544874488                - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`44894490                  Resolved default configuration for agent tools.44914492                  - `enabled: boolean`44934494                  - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`44954496                    Permission policy for tool execution.44974498                    - `BetaManagedAgentsAlwaysAllowPolicy object`44994500                      Tool calls are automatically approved without user confirmation.45014502                    - `BetaManagedAgentsAlwaysAskPolicy object`45034504                      Tool calls require user confirmation before execution.45054506                    - `BetaManagedAgentsAutoPolicy object`45074508                      The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.45094510              - `BetaManagedAgentsMCPToolset object`45114512                - `type: "mcp_toolset"`45134514                - `configs: array of BetaManagedAgentsMCPToolConfig`45154516                  - `enabled: boolean`45174518                  - `name: string`45194520                  - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`45214522                    Permission policy for tool execution.45234524                    - `BetaManagedAgentsAlwaysAllowPolicy object`45254526                      Tool calls are automatically approved without user confirmation.45274528                    - `BetaManagedAgentsAlwaysAskPolicy object`45294530                      Tool calls require user confirmation before execution.45314532                    - `BetaManagedAgentsAutoPolicy object`45334534                      The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.45354536                - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`45374538                  Resolved default configuration for all tools from an MCP server.45394540                  - `enabled: boolean`45414542                  - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`45434544                    Permission policy for tool execution.45454546                    - `BetaManagedAgentsAlwaysAllowPolicy object`45474548                      Tool calls are automatically approved without user confirmation.45494550                    - `BetaManagedAgentsAlwaysAskPolicy object`45514552                      Tool calls require user confirmation before execution.45534554                    - `BetaManagedAgentsAutoPolicy object`45554556                      The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.45574558                - `mcp_server_name: string`45594560              - `BetaManagedAgentsCustomTool object`45614562                A custom tool as returned in API responses.45634564                - `type: "custom"`45654566                - `description: string`45674568                - `input_schema: BetaManagedAgentsCustomToolInputSchema`45694570                  JSON Schema for custom tool input parameters.45714572                  - `type: "object"`45734574                  - `properties: optional map[unknown] or null`45754576                  - `required: optional array of string or null`45774578                - `name: string`45794580            - `version: number`45814582              format: int3245834584          - `BetaManagedAgentsAdvisor object`45854586            Platform advisor roster entry: a model the session's primary thread may consult mid-turn.45874588            - `type: "advisor"`45894590            - `model: string`45914592              The advisor model id.45934594      - `name: string`45954596      - `skills: array of BetaManagedAgentsAnthropicSkill or BetaManagedAgentsCustomSkill`45974598        - `BetaManagedAgentsAnthropicSkill object`45994600          A resolved Anthropic-managed skill.46014602        - `BetaManagedAgentsCustomSkill object`46034604          A resolved user-created custom skill.46054606      - `system: string or null`46074608      - `tools: array of BetaManagedAgentsAgentToolset20260401 or BetaManagedAgentsMCPToolset or BetaManagedAgentsCustomTool`46094610        - `BetaManagedAgentsAgentToolset20260401 object`46114612        - `BetaManagedAgentsMCPToolset object`46134614        - `BetaManagedAgentsCustomTool object`46154616          A custom tool as returned in API responses.46174618      - `version: number`46194620        format: int3246214622    - `budget: optional BetaManagedAgentsBudgetLimit or null`46234624      A hard spend ceiling. The session stops issuing new model requests once the tracked list cost reaches `max_list_cost`.46254626      - `type: "limit"`46274628      - `max_list_cost: BetaMonetaryAmount`46294630        A monetary amount in a specific currency.46314632        - `amount: string`46334634          Amount in minor units of the currency, as an integer decimal string with no leading zeros: "2500" is $25.00 and "50" is fifty cents. A string rather than a number so no float rounding is ever applied.46354636        - `currency: BetaCurrency`46374638          Uppercase ISO-4217 currency code. `USD` is the only currency currently supported; the accepted set is closed and grows only when a new currency is priced.46394640    - `metadata: optional map[string]`46414642      The session's full metadata bag after the update. Present when the update set non-empty metadata; absent when metadata was unchanged or cleared to empty.46434644    - `title: optional string or null`46454646      The session's new title. Present only when the update changed it.46474648  - `BetaManagedAgentsStartEvent object`46494650    Opens a preview of a buffered event. Carries the previewed event's type and id only. Followed by zero or more event_delta events with the same event id, normally concluded by the buffered event carrying that id. If the producing model request ends without that event (an error or interrupt mid-stream), its terminal span.model_request_end closes the preview. Only sent on stream connections that opt in via event_deltas; never appears in event history.46514652    - `type: "event_start"`46534654    - `event: BetaManagedAgentsStartEventPreview`46554656      The previewed event's type and id. The event type determines which delta types the preview's event_delta events carry: agent.message events stream content_delta fragments; agent.thinking previews are start-only — no deltas follow, and the buffered agent.thinking with the same id concludes them.46574658      - `BetaManagedAgentsAgentMessagePreview object`46594660        - `type: "agent.message"`46614662        - `id: string`46634664          The id the buffered agent.message will carry if it is emitted. Matches the event_id on this preview's event_delta events.46654666      - `BetaManagedAgentsAgentThinkingPreview object`46674668        - `type: "agent.thinking"`46694670        - `id: string`46714672          The id the buffered agent.thinking will carry if it is emitted. Start-only — no event_delta events follow.46734674  - `BetaManagedAgentsDeltaEvent object`46754676    An incremental update to an event that is still being streamed. Deltas are best-effort and may stop early; when the buffered event with id == event_id is produced it carries the complete content. A model request that ends early (an error or interrupt) produces no buffered event — its terminal span.model_request_end closes the preview. Only sent on stream connections that opt in via event_deltas; never appears in event history.46774678    - `type: "event_delta"`46794680    - `delta: BetaManagedAgentsDeltaContent`46814682      One fragment of the previewed event. The delta type is named for the previewed event's field it streams into: agent.message events stream content_delta fragments, each a partial element of the content array.46834684      - `type: "content_delta"`46854686      - `content: BetaManagedAgentsTextBlock`46874688        Regular text content.46894690      - `index: optional number`46914692        Which entry in the previewed event's content array this fragment lands in. Insert content as that entry when the index is new; append to the existing entry otherwise.46934694        format: uint3246954696    - `event_id: string`46974698      The id of the event being previewed. Matches event.id on the corresponding event_start and the buffered event that reconciles the preview.46994700  - `BetaManagedAgentsSystemMessageEvent object`47014702    A mid-conversation system message event. Carries system-role content that is appended to the session as a `role: "system"` turn.47034704    - `type: "system.message"`47054706    - `id: string`47074708      Unique identifier for this event.47094710    - `content: array of BetaManagedAgentsSystemContentBlock`47114712      System content blocks. Text-only.47134714      - `type: "text"`47154716      - `text: string`47174718        The text content.47194720        minLength: 147214722    - `processed_at: optional string or null`47234724      A timestamp in RFC 3339 format47254726      format: date-time47274728  - `BetaManagedAgentsSessionUsageEvent object`47294730    Periodic snapshot of the session's cumulative usage and tracked list cost.47314732    - `type: "session.usage"`47334734    - `id: string`47354736      Unique identifier for this event.47374738    - `processed_at: string`47394740      A timestamp in RFC 3339 format47414742      format: date-time47434744    - `usage: BetaManagedAgentsSessionUsageSnapshot`47454746      Point-in-time snapshot of a session's cumulative usage.47474748      - `active_seconds: optional number`47494750        Cumulative time in seconds during which the session had at least one thread in running status. Overlapping activity from concurrent threads is counted once. This is the duration the session's runtime cost is priced on.47514752        format: double47534754      - `cache_creation: optional BetaManagedAgentsCacheCreationUsage`47554756        Prompt-cache creation token usage broken down by cache lifetime.47574758        - `ephemeral_1h_input_tokens: optional number`47594760          Tokens used to create 1-hour ephemeral cache entries.47614762          format: int3247634764        - `ephemeral_5m_input_tokens: optional number`47654766          Tokens used to create 5-minute ephemeral cache entries.47674768          format: int3247694770      - `cache_read_input_tokens: optional number`47714772        Total tokens read from prompt cache.47734774        format: int3247754776      - `input_tokens: optional number`47774778        Total input tokens consumed across all turns.47794780        format: int3247814782      - `list_cost: optional BetaMonetaryAmount`47834784        A monetary amount in a specific currency.47854786      - `output_tokens: optional number`47874788        Total output tokens generated across all turns.47894790        format: int3247914792      - `server_tool_use: optional BetaManagedAgentsServerToolUsage`47934794        Cumulative count of server-executed tool invocations, broken down by tool.47954796        - `web_fetch_requests: optional number`47974798          Number of server-executed web fetch requests.47994800          format: int3248014802        - `web_search_requests: optional number`48034804          Number of server-executed web search requests.48054806          format: int3248074808    - `budget: optional BetaManagedAgentsBudgetLimit or null`48094810      A hard spend ceiling. The session stops issuing new model requests once the tracked list cost reaches `max_list_cost`.48114812### Example48134814```bash4815curl https://api.anthropic.com/v1/sessions/$SESSION_ID/threads/$THREAD_ID/stream \4816    -H 'anthropic-version: 2023-06-01' \4817    -H 'anthropic-beta: managed-agents-2026-04-01' \4818    -H "X-Api-Key: $ANTHROPIC_API_KEY"4819```48204821#### Response (200)48224823```json4824{4825  "id": "sevt_011CZkZGOp0iBcp4kaQSihUmy",4826  "content": [4827    {4828      "text": "Where is my order #1234?",4829      "type": "text"4830    }4831  ],4832  "type": "user.message",4833  "processed_at": "2026-03-15T10:00:00Z"4834}4835```4836