> ## Documentation Index > Fetch the complete documentation index at: https://docs.langchain.com/llms.txt > Use this file to discover all available pages before exploring further. # Get an agent > Return the specified agent, including metadata, permissions, runtime, extras, and the parsed file tree at the latest commit. Pass `include_files=true` to include the raw file map. ## OpenAPI ````yaml /langsmith/managed-deep-agents-openapi.json get /agents/{agent_id} openapi: 3.1.0 info: title: Managed Deep Agents API version: 0.1.0 description: Create, configure, and run Managed Deep Agents. servers: - url: https://api.smith.langchain.com/v1/deepagents security: - apiKeyAuth: [] tags: - name: Agents description: Create and manage Managed Deep Agent resources. - name: Threads description: Create and inspect durable thread state. - name: Runs description: Start runs on Managed Deep Agent threads. - name: MCP servers description: Register MCP servers and store credentials referenced by agent tools. - name: MCP tools description: List tools exposed by registered MCP servers. - name: Auth sessions description: Start and poll OAuth authorization sessions for OAuth MCP servers. paths: /agents/{agent_id}: get: tags: - Agents summary: Get an agent description: >- Return the specified agent, including metadata, permissions, runtime, extras, and the parsed file tree at the latest commit. Pass `include_files=true` to include the raw file map. operationId: getManagedDeepAgent parameters: - $ref: '#/components/parameters/AgentID' - $ref: '#/components/parameters/IncludeFiles' responses: '200': description: Agent returned. content: application/json: schema: $ref: '#/components/schemas/Agent' '400': $ref: '#/components/responses/BadRequest' '401': $ref: '#/components/responses/Unauthorized' '403': $ref: '#/components/responses/Forbidden' '404': $ref: '#/components/responses/NotFound' '500': $ref: '#/components/responses/InternalServerError' '502': $ref: '#/components/responses/BadGateway' '503': $ref: '#/components/responses/ServiceUnavailable' components: parameters: AgentID: name: agent_id in: path required: true description: Managed Deep Agent ID. schema: type: string format: uuid IncludeFiles: name: include_files in: query required: false description: >- Include the raw file map in the agent response. Accepted truthy values are `true` and `1`. schema: type: boolean default: false schemas: Agent: allOf: - $ref: '#/components/schemas/AgentSummary' - type: object properties: revision: type: string description: Revision token for the latest file tree commit. instructions: type: string description: Agent system prompt parsed from `AGENTS.md`. tools: $ref: '#/components/schemas/ToolsConfig' subagents: type: array items: $ref: '#/components/schemas/SubagentSpec' skills: type: array items: $ref: '#/components/schemas/SkillSpec' files: $ref: '#/components/schemas/FileMap' AgentSummary: type: object properties: id: type: string format: uuid name: type: string description: type: string owner_id: type: string format: uuid created_at: type: string format: date-time updated_at: type: string format: date-time permissions: $ref: '#/components/schemas/AgentPermissions' runtime: $ref: '#/components/schemas/AgentRuntime' backend: $ref: '#/components/schemas/AgentBackend' extras: type: object additionalProperties: true ToolsConfig: type: object properties: tools: type: array description: MCP tools available to the agent. Serialized to `tools.json`. items: $ref: '#/components/schemas/ToolSpec' interrupt_config: type: object description: >- Per-tool interrupt rules. Keys are tool identifiers in the form `{mcp_server_url}::{tool_name}` (separator is two colons). The URL is normalized with any trailing slash stripped, so `https://tools.example.com` and `https://tools.example.com/` match the same tool. Additional `::`-separated components after the tool name (such as the MCP server display name) are accepted but ignored when matching. A bare tool name with no `::` is also accepted as a fallback. Values are booleans: `true` requires a human-in-the-loop interrupt before the tool runs; `false` allows it without interrupt. Use `{}` for no interrupts. additionalProperties: type: boolean SubagentSpec: type: object properties: name: type: string description: >- Subagent slug. Must be non-blank and may contain letters, numbers, spaces, underscores, or hyphens. description: type: string description: Short human-readable description. model_id: type: string description: >- Model ID for this subagent. Inherits the parent runtime when empty. The REST API field is `model_id`; local CLI project files use `model`. instructions: type: string description: Subagent system prompt. Written to `subagents//AGENTS.md`. tools: $ref: '#/components/schemas/ToolsConfig' SkillSpec: type: object properties: type: type: string enum: - inline default: inline description: Skill discriminator. Only `inline` is supported. name: type: string description: >- Skill slug. Must be non-blank and may contain letters, numbers, spaces, underscores, or hyphens. description: type: string description: Short human-readable description shown to the agent. instructions: type: string description: Skill body. Written to `skills//SKILL.md`. files: type: object description: >- Supporting files keyed by relative path. Paths must be relative with no leading slash or `..` segment. additionalProperties: type: string FileMap: type: object description: >- Raw file map for paths not covered by typed fields. Keys are relative file paths. Setting a typed field and the corresponding `files` entry returns `422`. additionalProperties: $ref: '#/components/schemas/FileEntry' ErrorResponse: type: object properties: type: type: string code: type: string detail: type: string status: type: integer AgentPermissions: type: object description: >- Agent visibility and sharing configuration. When omitted on create, defaults to personal identity, tenant visibility, and read tenant access. properties: identity: type: string enum: - personal - shared default: personal description: Agent identity scope. visibility: type: string enum: - tenant - user default: tenant description: Visibility scope for the agent. tenant_access_level: type: string enum: - read - run - write default: read description: Workspace access level. shared_users: $ref: '#/components/schemas/SharedUsers' AgentRuntime: type: object properties: model: $ref: '#/components/schemas/AgentModelConfig' AgentBackend: type: object description: >- Backend configuration for the agent. Use `default` when the agent does not need sandbox-specific backend behavior. Use `thread_scoped_sandbox` or `agent_scoped_sandbox` when the agent needs LangSmith sandbox resources for code execution, filesystem work, or long-running tasks. See [Sandboxes overview](/langsmith/sandboxes). properties: type: type: string enum: - default - thread_scoped_sandbox - agent_scoped_sandbox description: >- Backend type. `thread_scoped_sandbox` scopes LangSmith sandbox resources to each thread. `agent_scoped_sandbox` scopes LangSmith sandbox resources to the agent. `default` uses no sandbox-specific backend behavior. example: default sandbox: $ref: '#/components/schemas/SandboxBackendConfig' ToolSpec: type: object properties: name: type: string description: Tool name as exposed by the MCP server. mcp_server_url: type: string format: uri description: URL of the MCP server hosting this tool. mcp_server_name: type: string description: Display name for the MCP server. Defaults to the URL when empty. display_name: type: string description: Display name for the tool. Defaults to the tool name when empty. FileEntry: type: object properties: content: type: string description: File contents as a UTF-8 string. SharedUsers: type: object properties: read: type: array description: User IDs allowed to read the agent. items: type: string run: type: array description: User IDs allowed to run the agent. items: type: string update: type: array description: User IDs allowed to update the agent. items: type: string AgentModelConfig: type: object properties: model_id: type: string description: >- Model identifier used to run the agent. Use the form `{provider}:{model_name}`, for example, `openai:gpt-5.5`. Values without a colon are interpreted as references to a saved Playground configuration, not as direct model identifiers. See [Supported models](/langsmith/managed-deep-agents-overview#supported-models) for the recognized provider prefixes. example: openai:gpt-5.5 SandboxBackendConfig: type: object description: >- LangSmith sandbox settings for `thread_scoped_sandbox` and `agent_scoped_sandbox` backends. properties: policy_ids: type: array description: Sandbox policy IDs. items: type: string idle_ttl_seconds: type: integer description: Sandbox idle timeout in seconds. delete_after_stop_seconds: type: integer description: Sandbox deletion delay after stop, in seconds. responses: BadRequest: description: Bad request. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' examples: missingRequiredField: summary: Required field missing value: type: https://docs.langchain.com/errors/missing_required_field code: missing_required_field detail: 'Set ''name'' to a non-empty string: ''name'' is required' status: 400 invalidRequestBody: summary: Malformed JSON body value: type: https://docs.langchain.com/errors/invalid_request_body code: invalid_request_body detail: >- Provide a valid JSON request body matching the documented schema status: 400 Unauthorized: description: >- Authentication required. The auth layer returns a simple `{"error": "Unauthorized"}` body that does not match the structured `ErrorResponse` shape used by other 4xx responses. content: application/json: schema: type: object properties: error: type: string examples: missingApiKey: summary: No `X-Api-Key` header supplied value: error: Unauthorized Forbidden: description: >- The authenticated identity does not have access to the resource or feature. The auth layer returns a simple `{"error": "Forbidden"}` body that does not match the structured `ErrorResponse` shape used by other 4xx responses. content: application/json: schema: type: object properties: error: type: string examples: invalidApiKey: summary: API key is invalid or lacks workspace access value: error: Forbidden NotFound: description: Resource not found. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' examples: agentNotFound: summary: Unknown agent ID value: type: https://docs.langchain.com/errors/agent_not_found code: agent_not_found detail: >- Verify the agent ID and that the calling identity has access: agent not found status: 404 InternalServerError: description: Internal server error. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' BadGateway: description: Upstream agent runtime error. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' examples: upstreamAuthFailed: summary: >- Upstream runtime rejected the request (e.g. orphaned thread after agent deletion) value: type: https://docs.langchain.com/errors/upstream_auth_failed code: upstream_auth_failed detail: >- Verify that the calling identity is authorized for this organization: upstream agent runtime rejected the request status: 502 ServiceUnavailable: description: Dependent service unavailable. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' securitySchemes: apiKeyAuth: type: apiKey in: header name: X-Api-Key x-default: LANGSMITH_API_KEY ````