> ## Documentation Index > Fetch the complete documentation index at: https://docs.duvo.ai/llms.txt > Use this file to discover all available pages before exploring further. # Create Agent > Create a new agent. Optionally include a build configuration to create the first build in the same request. On a team that versions its automations, an agent created inside an automation joins that automation's draft revision: its first build is a draft and goes live when the draft is made active. ## OpenAPI ````yaml https://api.duvo.ai/v2/documentation/mintlify.json post /v2/teams/{teamId}/agents openapi: 3.0.3 info: title: Duvo Public API description: >- Public API for programmatic access to Duvo. Authenticate with API keys created in the Duvo dashboard. ## Rate limits Requests are counted per API key. Every response advertises the quota with the IETF RateLimit fields, so a client can self-throttle without waiting for a rejection: - `RateLimit-Policy: "default";q=;w=` — the policy in force: `q` requests per `w` seconds. - `RateLimit: "default";r=;t=` — the live state: `r` requests left, quota resetting in `t` seconds. - Read the quota from the headers rather than hard-coding it; it differs per environment. - The same numbers are also sent as `ratelimit-limit`, `ratelimit-remaining` and `ratelimit-reset`. Once the quota is exhausted the API answers `429 Too Many Requests` with `Retry-After` set to the seconds to wait; honor it instead of retrying immediately. version: 1.0.0 servers: - url: https://api.duvo.ai description: Production server security: [] tags: - name: Runs description: Start, monitor, and manage agent runs (Runs in the Duvo UI) - name: Sandboxes description: Create sandboxes and upload files for agent runs - name: Queues description: Manage queues and their agent bindings - name: Cases description: Create, list, and manage cases and their labels within queues - name: Case Approvals description: Submit decisions on pending case approval requests issued by an agent run - name: Case Attachments description: Upload, list, download, and remove the files attached to a case - name: Automations description: >- List and manage automations — the workspace container that groups the agents and queues making up one end-to-end process - name: Agents description: Create and manage agents for automation workloads - name: Revisions description: Create and manage agent revisions — the underlying Setup for an Agent - name: Agent Folders description: Organize agents into folders - name: Agent Memory description: Read an agent's memory files (the Memory feature in the Duvo UI) - name: Suggestions description: >- List, apply, and dismiss an Agent's improvement suggestions (the suggestions inbox in the Duvo UI) - name: Notifications description: >- List, read, dismiss, and clear the authenticated user's team notifications (the Notification Center in the Duvo UI) - name: Schedules description: List schedules configured for an agent - name: Duvo Pulse description: >- Create, list, iterate on, and delete Duvo Pulse dashboards — live, agent-generated visualizations of your Duvo data - name: Case Triggers description: >- Configure case triggers that automatically dispatch agent runs (Runs in the Duvo UI) for cases added to a queue - name: Triggers description: >- Configure event triggers that start a Run automatically when an external event fires (e.g. an email arrives, a Linear issue is created, or a file changes in Google Drive) - name: Skills description: Manage team and system skills (reusable knowledge packs). - name: Files description: Manage team files. - name: Plugins description: Discover plugins that can be referenced from a revision. - name: Organizations description: Inspect organizations you belong to and the teams within them - name: Team description: Inspect the team and members associated with the API key - name: Invites description: >- Invite people to a team — one at a time or in bulk, scoped to a Clarity process or the whole team — and manage the team's shareable invite link - name: Integrations description: Browse the team's catalog of available integration types - name: Connections description: Manage your connected integrations - name: Credentials description: >- Manage logins (domain + username + password + TOTP) used by agents to sign in to websites and desktop applications, and attach them to assignment revisions - name: Secrets description: >- Manage env-var secrets injected into runs, and attach them to assignment revisions. Only metadata is exposed; values are never returned - name: Revision Integrations description: >- Attach integrations to assignment revisions, pin specific connections, and link queues - name: ClarityV2 description: >- Manage Clarity v2 process snapshots, automation proposals, and the extra-capture-request follow-up loop paths: /v2/teams/{teamId}/agents: post: tags: - Agents summary: Create Agent description: >- Create a new agent. Optionally include a build configuration to create the first build in the same request. On a team that versions its automations, an agent created inside an automation joins that automation's draft revision: its first build is a draft and goes live when the draft is made active. operationId: createAgent parameters: - name: teamId in: path required: true schema: type: string requestBody: required: true content: application/json: schema: type: object properties: name: type: string minLength: 1 description: Human-readable agent name type: default: agent description: >- Flavour of the step to create. Fixed at creation — it selects the config family every build of this step must use, and the two cannot be mixed. type: string enum: - agent - code thread_id: nullable: true description: Existing thread ID to associate with the agent type: string format: uuid source_build_id: nullable: true description: >- Build ID whose case-queue-consumer setup should be copied to the new agent. When `automation_id` is omitted, the new agent also joins the automation of the agent this build belongs to. type: string format: uuid automation_id: nullable: true description: >- Automation this agent belongs to. If omitted on a team using the automations model, a new automation named after the agent is created automatically. type: string format: uuid build: type: object properties: name: type: string minLength: 1 description: Build name config: type: object properties: version: type: string enum: - v2 description: >- Schema version discriminator — must be "v2" for the current schema data: type: object properties: models: type: object properties: agent: type: object properties: model: type: string x-extensible-enum: - claude-haiku-4-5-20251001 - claude-sonnet-5 - claude-sonnet-5[1m] - claude-sonnet-5-5 - claude-sonnet-5-5[1m] - claude-opus-5 - claude-opus-5[1m] - claude-opus-5-5 - claude-opus-5-5[1m] - kimi-k3 - kimi-k3-duvo - glm-5.3-flash - deepseek-v4-flash - minimax-m3 - d1-max - duvo-1-max - duvo-1-max-sonnet-1m - duvo-1-max-sonnet-4.5 - duvo-1-max-sonnet-4.5-1m - duvo-1-max-opus - duvo-1-max-opus-4.5 - gpt-4.1 - gpt-4o - gpt-4o-mini - gpt-5 - gpt-5.1 - claude-sonnet-4-20250514 - claude-sonnet-4-20250514[1m] - claude-sonnet-4-5-20250929 - claude-sonnet-4-5-20250929[1m] - claude-sonnet-4-6 - claude-sonnet-4-6[1m] - claude-opus-4-1-20250805 - claude-opus-4-5-20251101 - claude-opus-4-6 - claude-opus-4-6[1m] - claude-opus-4-7 - claude-opus-4-7[1m] - claude-opus-4-8 - claude-opus-4-8[1m] - glm-5.2 - qwen3.6-27b - qwen3.8-max description: >- Model identifier used for the primary agent loop (Claude, or an OSS model when the team's oss_models flag — or oss_models_public for the public subset — is enabled) required: - model description: Primary agent model configuration browsing: type: object properties: provider: type: string enum: - google - anthropic description: >- Provider backing the browsing/computer-use model model: type: string enum: - gemini-2.5-pro - gemini-2.5-flash - gemini-3-pro-preview - claude-haiku-4-5 - claude-sonnet-4-5 - claude-opus-4-1 description: >- Model identifier for the browsing provider required: - provider - model description: Browsing/computer-use model configuration required: - agent - browsing description: >- Model configuration for each capability the agent uses input: anyOf: - type: string - type: array items: type: object properties: role: type: string enum: - system - user - assistant content: anyOf: - type: string - type: array items: anyOf: - type: object properties: type: type: string enum: - input_text text: type: string required: - type - text - type: object properties: type: type: string enum: - input_image detail: nullable: true type: string enum: - low - high - auto file_id: type: string nullable: true url: type: string nullable: true required: - type - type: object properties: type: type: string enum: - input_file file_id: type: string required: - type - file_id type: nullable: true type: string enum: - message required: - role - content description: >- Initial agent instructions — either a single system prompt string or a list of structured messages files: default: [] description: >- Team files that should be available to the agent, each given as the relative file path returned in the `path` field of `GET /v2/teams/{teamId}/files` (for example `report.md` or `folder/doc.md`) — not the `id` field, which is a GCS object identifier. nullable: true type: array items: type: string skills: default: [] description: >- IDs of skills (team or system) that should be available to the agent nullable: true type: array items: type: string plugins: default: [] description: >- Plugins to load — either a built-in plugin name (e.g. "code-review") or a GitHub URL (e.g. "https://github.com/owner/repo") nullable: true type: array items: type: string subAgents: default: [] description: >- Retired — Sub-Assignments was removed. Accepted for backwards compatibility and ignored. nullable: true type: array items: type: string format: uuid options: description: >- Optional runtime options controlling how the agent executes type: object properties: browserProvider: description: Browser infrastructure provider type: string enum: - browserbase - browser-use evaluationSchemaId: description: >- ID of the evaluation schema to apply to runs of this agent type: string benchmarkExpectedOutcomes: description: >- Expected outcomes used when running benchmark scenarios type: array items: type: object properties: description: type: string description: >- Human-readable description of the expected outcome criteria: type: string description: >- Pass/fail criteria used to evaluate the outcome required: - description - criteria supervisedMode: description: >- Run the agent under Duvo's permission classifier: every tool call is classified as allow/deny/ask, with 'ask' routed to human-in-the-loop approval. Mutually exclusive with browsing and computer-use connections. type: boolean additionalProperties: {} required: - models - input description: Agent configuration payload required: - version - data description: >- Build configuration (latest schema version only; legacy v1 configs are not accepted). Agent steps only — a code step's first build is created via `POST /v2/agents/{agent_id}/revisions`, which accepts a code-step config. required: - name - config description: >- Optional inline build. When present, a first build is created alongside the agent. Only valid for an agent step. required: - name responses: '201': description: Default Response content: application/json: schema: type: object properties: agent: type: object properties: id: type: string format: uuid name: type: string type: description: >- Flavour of the step: "agent" for an agentic step, "code" for a deterministic code step. Fixed at creation. type: string enum: - agent - code slack_enabled: type: boolean nullable: true microsoft_teams_enabled: type: boolean nullable: true agentic_memory_enabled: type: boolean nullable: true computer_use_sandbox_template_id: nullable: true type: string description: >- "standard" when durably pinned to the standard Duvo desktop, a template UUID when pinned to a specific desktop, or null when following the team default. computer_use_vpn_config_id: nullable: true type: string description: >- "direct" when durably pinned to direct internet, a VPN config UUID when pinned to a specific network, or null when following the team default. team_id: type: string nullable: true thread_id: type: string nullable: true folder_id: type: string nullable: true automation_id: nullable: true type: string format: uuid created_at: type: string nullable: true updated_at: type: string nullable: true created_by: nullable: true type: object properties: id: type: string name: type: string nullable: true email: type: string required: - id - name - email additionalProperties: false updated_by: nullable: true type: object properties: id: type: string name: type: string nullable: true email: type: string required: - id - name - email additionalProperties: false required: - id - name - slack_enabled - microsoft_teams_enabled - agentic_memory_enabled - team_id - thread_id - created_at - updated_at additionalProperties: false build: type: object properties: id: type: string format: uuid copilot_id: type: string format: uuid agent_id: type: string format: uuid name: type: string revision_name: type: string nullable: true revision_description: type: string nullable: true config: nullable: true anyOf: - type: object properties: version: type: string enum: - v1 description: >- Schema version discriminator — must be "v1" for the current code-step schema data: type: object properties: language: type: string enum: - python description: Language the step's program is written in source: oneOf: - type: object properties: kind: type: string enum: - inline code: type: string minLength: 1 description: >- Program held in the build config. Written to `main.py` and run with `python main.py`. required: - kind - code additionalProperties: false - type: object properties: kind: type: string enum: - files startCommand: type: string minLength: 1 maxLength: 2000 description: >- Command run from the root of the step's file system, e.g. "python main.py". bundle: description: >- The program's files, as a content-addressed manifest type: object properties: bucket: type: string minLength: 1 description: >- GCS bucket the bundle's blobs are stored in files: maxItems: 200 type: array items: type: object properties: path: type: string minLength: 1 description: >- Path the file is materialised at, relative to the work dir sha256: type: string minLength: 64 maxLength: 64 pattern: ^[0-9a-f]{64}$ description: >- Digest of the file's bytes, and the key its blob is stored at sizeBytes: type: integer minimum: 0 maximum: 131072 description: UTF-8 byte length of the file's contents required: - path - sha256 - sizeBytes additionalProperties: false description: >- The program's files, one manifest entry each required: - bucket - files additionalProperties: false required: - kind - startCommand additionalProperties: false description: Where the step's program comes from files: default: [] description: >- Team file paths this step can read at run time nullable: true type: array items: type: string timeoutMs: default: 600000 description: >- Wall-clock budget for one execution of the step type: integer exclusiveMinimum: true maximum: 3600000 handoverRequired: default: false description: >- When the step declares handover options, whether a run that exits cleanly without requesting one fails. Ignored when the step declares no options. type: boolean required: - language - source - files - timeoutMs - handoverRequired additionalProperties: false description: Code step configuration payload required: - version - data additionalProperties: false - type: object properties: version: type: string enum: - v1 data: type: object properties: model: type: string temperature: nullable: true type: number minimum: 0 maximum: 2 max_output_tokens: nullable: true type: number minimum: 0 exclusiveMinimum: true top_p: nullable: true type: number minimum: 0 maximum: 1 store: type: boolean nullable: true tools: {} input: anyOf: - type: string - type: array items: type: object properties: role: type: string enum: - system - user - assistant content: anyOf: - type: string - type: array items: anyOf: - type: object properties: type: type: string enum: - input_text text: type: string required: - type - text additionalProperties: false - type: object properties: type: type: string enum: - input_image detail: nullable: true type: string enum: - low - high - auto file_id: type: string nullable: true url: type: string nullable: true required: - type additionalProperties: false - type: object properties: type: type: string enum: - input_file file_id: type: string required: - type - file_id additionalProperties: false type: nullable: true type: string enum: - message required: - role - content additionalProperties: false text: nullable: true type: object properties: format: nullable: true anyOf: - type: object properties: type: type: string enum: - text required: - type additionalProperties: false - type: object properties: type: type: string enum: - json_object required: - type additionalProperties: false additionalProperties: false files: default: [] nullable: true type: array items: type: string skills: default: [] nullable: true type: array items: type: string required: - model - tools - input - files - skills additionalProperties: false required: - version - data additionalProperties: false - type: object properties: version: type: string enum: - v2 description: >- Schema version discriminator — must be "v2" for the current schema data: type: object properties: models: type: object properties: agent: type: object properties: model: type: string x-extensible-enum: - claude-haiku-4-5-20251001 - claude-sonnet-5 - claude-sonnet-5[1m] - claude-sonnet-5-5 - claude-sonnet-5-5[1m] - claude-opus-5 - claude-opus-5[1m] - claude-opus-5-5 - claude-opus-5-5[1m] - kimi-k3 - kimi-k3-duvo - glm-5.3-flash - deepseek-v4-flash - minimax-m3 - d1-max - duvo-1-max - duvo-1-max-sonnet-1m - duvo-1-max-sonnet-4.5 - duvo-1-max-sonnet-4.5-1m - duvo-1-max-opus - duvo-1-max-opus-4.5 - gpt-4.1 - gpt-4o - gpt-4o-mini - gpt-5 - gpt-5.1 - claude-sonnet-4-20250514 - claude-sonnet-4-20250514[1m] - claude-sonnet-4-5-20250929 - claude-sonnet-4-5-20250929[1m] - claude-sonnet-4-6 - claude-sonnet-4-6[1m] - claude-opus-4-1-20250805 - claude-opus-4-5-20251101 - claude-opus-4-6 - claude-opus-4-6[1m] - claude-opus-4-7 - claude-opus-4-7[1m] - claude-opus-4-8 - claude-opus-4-8[1m] - glm-5.2 - qwen3.6-27b - qwen3.8-max description: >- Model identifier used for the primary agent loop (Claude, or an OSS model when the team's oss_models flag — or oss_models_public for the public subset — is enabled) required: - model additionalProperties: false description: Primary agent model configuration browsing: type: object properties: provider: type: string enum: - google - anthropic description: >- Provider backing the browsing/computer-use model model: type: string enum: - gemini-2.5-pro - gemini-2.5-flash - gemini-3-pro-preview - claude-haiku-4-5 - claude-sonnet-4-5 - claude-opus-4-1 description: >- Model identifier for the browsing provider required: - provider - model additionalProperties: false description: >- Browsing/computer-use model configuration required: - agent - browsing additionalProperties: false description: >- Model configuration for each capability the agent uses input: anyOf: - type: string - type: array items: type: object properties: role: type: string enum: - system - user - assistant content: anyOf: - type: string - type: array items: anyOf: - type: object properties: type: type: string enum: - input_text text: type: string required: - type - text additionalProperties: false - type: object properties: type: type: string enum: - input_image detail: nullable: true type: string enum: - low - high - auto file_id: type: string nullable: true url: type: string nullable: true required: - type additionalProperties: false - type: object properties: type: type: string enum: - input_file file_id: type: string required: - type - file_id additionalProperties: false type: nullable: true type: string enum: - message required: - role - content additionalProperties: false description: >- Initial agent instructions — either a single system prompt string or a list of structured messages files: default: [] description: >- Team files that should be available to the agent, each given as the relative file path returned in the `path` field of `GET /v2/teams/{teamId}/files` (for example `report.md` or `folder/doc.md`) — not the `id` field, which is a GCS object identifier. nullable: true type: array items: type: string skills: default: [] description: >- IDs of skills (team or system) that should be available to the agent nullable: true type: array items: type: string plugins: default: [] description: >- Plugins to load — either a built-in plugin name (e.g. "code-review") or a GitHub URL (e.g. "https://github.com/owner/repo") nullable: true type: array items: type: string subAgents: default: [] description: >- Retired — Sub-Assignments was removed. Accepted for backwards compatibility and ignored. nullable: true type: array items: type: string format: uuid options: description: >- Optional runtime options controlling how the agent executes type: object properties: browserProvider: description: Browser infrastructure provider type: string enum: - browserbase - browser-use evaluationSchemaId: description: >- ID of the evaluation schema to apply to runs of this agent type: string benchmarkExpectedOutcomes: description: >- Expected outcomes used when running benchmark scenarios type: array items: type: object properties: description: type: string description: >- Human-readable description of the expected outcome criteria: type: string description: >- Pass/fail criteria used to evaluate the outcome required: - description - criteria additionalProperties: false supervisedMode: description: >- Run the agent under Duvo's permission classifier: every tool call is classified as allow/deny/ask, with 'ask' routed to human-in-the-loop approval. Mutually exclusive with browsing and computer-use connections. type: boolean additionalProperties: {} required: - models - input - files - skills - plugins - subAgents additionalProperties: false description: Agent configuration payload required: - version - data additionalProperties: false status: type: string nullable: true revision_number: nullable: true type: number parent_build_id: type: string nullable: true created_by: type: string nullable: true created_by_name: type: string nullable: true created_at: type: string nullable: true updated_at: type: string nullable: true schedules: type: array items: type: object properties: id: type: string build_id: type: string agent_id: type: string user_id: type: string enabled: type: boolean frequency: type: string enum: - every_5_minutes - every_15_minutes - hourly - daily - workday - weekly - monthly - custom time: type: string nullable: true timezone: type: string day: type: string nullable: true day_of_month: nullable: true type: number cron: type: string nullable: true source: type: string nullable: true task: type: string nullable: true recurring: type: boolean nullable: true auto_disabled_at: type: string nullable: true auto_disabled_reason: type: string nullable: true consecutive_missing_connection_failures: type: integer minimum: 0 maximum: 9007199254740991 first_missing_connection_failure_at: type: string nullable: true last_missing_connection_failure_at: type: string nullable: true created_at: type: string updated_at: type: string user_name: type: string nullable: true user_email: type: string nullable: true required: - id - build_id - agent_id - user_id - enabled - frequency - time - timezone - day - day_of_month - cron - source - task - recurring - auto_disabled_at - auto_disabled_reason - consecutive_missing_connection_failures - first_missing_connection_failure_at - last_missing_connection_failure_at - created_at - updated_at additionalProperties: false handover_to: nullable: true type: string format: uuid handover_options: type: array items: type: object properties: id: type: string format: uuid name: type: string required: - id - name additionalProperties: false required: - id - copilot_id - agent_id - name - config - status - revision_number - parent_build_id - created_by - created_at - updated_at additionalProperties: false required: - agent additionalProperties: false '400': description: Default Response content: application/json: schema: type: object properties: error: type: string message: type: string required: - error additionalProperties: false '401': description: Default Response content: application/json: schema: type: object properties: error: type: string message: type: string required: - error additionalProperties: false '403': description: Default Response content: application/json: schema: type: object properties: error: type: string message: type: string required: - error additionalProperties: false '409': description: Default Response content: application/json: schema: type: object properties: error: type: string message: type: string required: - error additionalProperties: false '500': description: Default Response content: application/json: schema: type: object properties: error: type: string message: type: string required: - error additionalProperties: false security: - bearerAuth: [] components: securitySchemes: bearerAuth: type: http scheme: bearer description: API key authentication. Get your API key from the Duvo dashboard. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.