> ## Documentation Index > Fetch the complete documentation index at: https://docs.revoengine.com/llms.txt > Use this file to discover all available pages before exploring further. # Create an Agent > Create an Agent. The request enforces Agent roles, ownership, workspace boundaries, and lifecycle state. ## OpenAPI ````yaml /api-reference/openapi.json post /api/v1/agents openapi: 3.0.0 info: title: RevoEngine Platform API description: Official RevoEngine Platform API documentation. version: 1.3.0 contact: name: Support email: support@revoengine.com termsOfService: https://docs.revoengine.com/tos servers: - url: https://api.revoengine.com description: Production security: [] tags: - name: Accounts description: Users, service accounts, API keys, status, and platform access. externalDocs: description: Identity and access guide url: https://docs.revoengine.com/operate/identity-and-access - name: Activity description: Snapshot-backed activity and version history. externalDocs: description: Observability guide url: https://docs.revoengine.com/operate/observability - name: Agent Memory description: Administrative access to durable Agent memory. externalDocs: description: Agent memory and workspaces guide url: https://docs.revoengine.com/ai/memory-and-workspaces - name: Agent Plugins description: Agent plugin definitions, synchronization, testing, and lifecycle. externalDocs: description: Agent plugins guide url: https://docs.revoengine.com/ai/plugins - name: Agents description: >- Agent definitions, runs, sessions, inboxes, memory, tools, workspaces, and lifecycle. externalDocs: description: Autonomous Agents guide url: https://docs.revoengine.com/ai/agents - name: Assistant description: >- Assistant threads, messages, responses, goals, reports, sharing, attachments, and control. externalDocs: description: Assistant guide url: https://docs.revoengine.com/ai/assistant - name: Automation description: >- Jobs, templates, schedules, events, webhooks, execution history, and lifecycle. externalDocs: description: Automation guide url: https://docs.revoengine.com/operate/automation - name: Chat description: Collaborative chat threads, messages, reactions, uploads, and attachments. externalDocs: description: Team Chat guide url: https://docs.revoengine.com/operate/chat - name: Component deployments description: >- Custom Node.js runtime deployment attempts, readiness, activation, and rollback. externalDocs: description: Custom Node.js guide url: https://docs.revoengine.com/build/custom-nodejs - name: Components description: Components, elements, versions, validation, activation, and lifecycle. externalDocs: description: Components guide url: https://docs.revoengine.com/build/components - name: Config description: Instance configuration, usage, logs, and cache administration. externalDocs: description: Instance settings guide url: https://docs.revoengine.com/platform/instance-settings - name: Database Cache description: Instance Database cache values and statistics. externalDocs: description: Database Cache guide url: https://docs.revoengine.com/operate/database-cache - name: Database Views description: Database Views, previews, refreshes, exports, and controlled data actions. externalDocs: description: Database Views guide url: https://docs.revoengine.com/operate/database-views - name: Databases description: Database definitions, data, audits, exports, cloning, and lifecycle. externalDocs: description: Databases guide url: https://docs.revoengine.com/operate/databases - name: Endpoints description: >- Endpoint definitions, activation, validation, generated OpenAPI, and statistics. externalDocs: description: Endpoints guide url: https://docs.revoengine.com/operate/endpoints - name: Files description: Compatibility file listing, preview, download, and deletion operations. externalDocs: description: Files compatibility guide url: https://docs.revoengine.com/operate/files - name: Groups description: Groups, membership, avatars, and lifecycle. externalDocs: description: Groups guide url: https://docs.revoengine.com/operate/groups - name: Me description: Current user profile, keys, avatar, and session information. externalDocs: description: Authentication guide url: https://docs.revoengine.com/developers/authentication - name: Preferences description: User and instance preference values. externalDocs: description: Platform UI guide url: https://docs.revoengine.com/platform/ui-tour - name: Role Groups description: Role groups, members, assigned roles, and lifecycle. externalDocs: description: Permissions guide url: https://docs.revoengine.com/operate/permissions - name: Roles description: Available platform roles and role details. externalDocs: description: Permissions guide url: https://docs.revoengine.com/operate/permissions - name: Search description: Cross-domain search over objects visible to the caller. externalDocs: description: Search and discovery guide url: https://docs.revoengine.com/platform/search - name: Secrets description: >- Secret definitions, revisions, activation, reveal, disable, and destruction. externalDocs: description: Secrets guide url: https://docs.revoengine.com/operate/secrets - name: Storage description: >- Storage workspaces, folders, objects, upload sessions, signed access, and lifecycle. externalDocs: description: Storage guide url: https://docs.revoengine.com/operate/storage externalDocs: description: Documentation url: https://docs.revoengine.com paths: /api/v1/agents: post: tags: - Agents summary: Create an Agent description: >- Create an Agent. The request enforces Agent roles, ownership, workspace boundaries, and lifecycle state. operationId: AgentController_createAgent parameters: [] requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/CreateAgentDto' responses: '200': description: '' content: application/json: schema: $ref: '#/components/schemas/AgentPublicDto' security: - bearer: [] components: schemas: CreateAgentDto: type: object properties: name: type: string description: Stable display name of the agent. example: On-call incident triage minLength: 1 desc: type: string description: Optional operator-facing description. example: Investigates production alerts and prepares escalation notes. serviceAccountUserId: type: string description: User id of the service account used as the execution principal. format: uuid example: 11111111-1111-4111-8111-111111111111 ownerId: type: string description: Owning user or group id used for visibility and management. format: uuid example: 22222222-2222-4222-8222-222222222222 ownerType: type: string description: Owner type, typically USER or GROUP. example: USER metadata: type: object description: >- Optional resource metadata. Values may use any JSON type. Maximum 16 KiB and 64 top-level keys. Top-level keys starting with "__" are reserved for backend use. example: costCenter: ops config: description: >- Persisted Agent identity and runtime defaults, including work mode, persona, instructions, memory policy, and execution limits. example: model: configured-model reasoningEffort: medium workMode: operator persona: pragmatic maxActiveRootRuns: 2 maxChildRuns: 4 maxConcurrentChildRuns: 2 maxTicks: 80 allOf: - $ref: '#/components/schemas/AgentConfigDto' policy: description: >- Persisted policy constraints such as tool allowlists, agent plugins, and delegation ceilings. Runtime-only route metadata is not configured here. example: toolNames: - slack.send_message - storage.read maxChildRuns: 4 maxConcurrentChildRuns: 2 allOf: - $ref: '#/components/schemas/AgentPolicyDto' required: - name - serviceAccountUserId AgentPublicDto: type: object properties: agentId: type: string description: Stable agent id. example: 99999999-9999-4999-8999-999999999999 name: type: string description: Agent display name. example: Release Reporter status: type: string description: Agent lifecycle status. enum: - ACTIVE - PAUSED - DISABLED example: ACTIVE desc: type: string description: Operator-facing description. example: Summarizes release status and creates operator updates. ownerId: type: string description: Owning user or group id. example: 22222222-2222-4222-8222-222222222222 ownerType: type: string description: Owner type, typically USER or GROUP. example: USER serviceAccountUserId: type: string description: Service account user id used as execution principal. example: 11111111-1111-4111-8111-111111111111 rootStorageEntryId: type: string description: Agent workspace root storage entry id. example: 33333333-3333-4333-8333-333333333333 config: type: object description: Persisted agent runtime defaults and identity inputs. example: model: configured-model maxTicks: 24 policy: type: object description: Persisted execution constraints and role gates. example: maxTicks: 24 toolNames: - ref_search metadata: type: object description: Resource metadata. example: team: operations avatar: type: string description: Signed avatar URL, when an avatar is configured. example: https://storage.example/avatar.png summary: description: >- Per-agent operational summary when requested by list or overview endpoints. example: activeRunCount: 1 queuedInboxCount: 3 allOf: - $ref: '#/components/schemas/AgentSummaryDto' createdBy: type: string description: User id that created the agent. example: 44444444-4444-4444-8444-444444444444 createdAt: format: date-time type: string description: Creation timestamp. example: '2026-06-13T10:00:00.000Z' lastUpdateAt: format: date-time type: string description: Last update timestamp. example: '2026-06-13T10:15:00.000Z' lastUpdateBy: type: string description: User id that last updated the agent. example: 44444444-4444-4444-8444-444444444444 version: type: number description: Optimistic locking version. example: 3 required: - agentId - name - status - serviceAccountUserId - config - policy - createdBy - createdAt - lastUpdateAt - lastUpdateBy - version AgentConfigDto: type: object properties: model: type: string description: Optional execution model identifier enabled for this instance. minLength: 1 reasoningEffort: type: string description: Reasoning effort passed to assistant threads. enum: - low - medium - high - xhigh executionMode: type: string description: Default execution mode used for Agent turns. enum: - interactive - execute - background - automation planningPolicy: type: string description: >- Durable planning policy default used for agent-loop turns. standard executes normally; review_required creates a reviewed plan before interactive execution. enum: - standard - review_required workMode: type: string description: >- Operating mode used to tune the agent toward the selected kind of work. enum: - balanced - coding - business - analyst - operator - creative persona: type: string description: >- Response persona applied beneath runtime policy and explicit agent instructions. enum: - friendly - pragmatic - professional - concise - calm - mentor - assertive systemInstructions: type: string description: >- Additional per-agent system instructions persisted with the definition. Runtime policy still has higher priority. minLength: 1 definition: description: >- Structured operator-owned Agent identity loaded on every autonomous turn. Learned memory cannot mutate this definition. allOf: - $ref: '#/components/schemas/AgentDefinitionDto' memoryPolicy: description: >- Durable cross-run memory defaults such as recall budget, retained summary size, and retention horizon. allOf: - $ref: '#/components/schemas/AgentMemoryPolicyDto' release: description: Release metadata for the current Agent definition. allOf: - $ref: '#/components/schemas/AgentReleaseConfigDto' automation: description: >- Enterprise automation defaults such as delivery mode, approval routing, and future delivery settings. allOf: - $ref: '#/components/schemas/AgentAutomationPolicyConfigDto' toolNames: description: >- Additional registered tool names associated with this agent. The runtime still exposes the complete callable tool registry. type: array items: type: string pluginIds: description: >- Default registry plugin ids attached to this agent. These are the agent-owned baseline attachments applied to every run unless additional runtime plugin ids are supplied. type: array items: type: string format: uuid plugins: description: >- Primary plugin contract surface. Plugins are provider-governed and define the executable agent-plugin surface. type: array items: $ref: '#/components/schemas/AgentPluginConfigDto' maxActiveRootRuns: type: number description: >- Maximum number of concurrently admitted root runs for this agent. Additional root runs remain queued until capacity is released. minimum: 1 maximum: 32 default: 1 maxChildRuns: type: number description: >- Maximum number of delegated child runs the parent may spawn per execution plan. minimum: 1 maximum: 128 maxConcurrentChildRuns: type: number description: >- Maximum number of delegated child runs that one parent may start concurrently. Remaining delegated steps wait for a later parent tick. minimum: 1 maximum: 32 default: 4 maxTicks: type: number description: Maximum number of ticks allowed for one run. minimum: 1 maximum: 1000 childFailureMode: type: string description: How the parent reacts when one delegated child fails. enum: - fail_parent - merge_partial AgentPolicyDto: type: object properties: defaultToolPermission: type: string description: >- Default approval mode applied to agent tool calls when no per-tool override exists. enum: - default - auto_review - full toolPermissions: type: object description: Per-tool approval mode overrides keyed by the runtime tool name. additionalProperties: enum: - default - auto_review - full toolNames: description: >- Additional policy-managed tool names. Tool permissions govern operations without truncating the callable registry. type: array items: type: string pluginIds: description: >- Policy-enforced registry plugin ids attached to this agent. These are governance-managed attachments and are merged in regardless of run-level plugin selection. type: array items: type: string format: uuid plugins: description: >- Policy-managed plugins merged with config plugins before execution-time routing narrows the active surface. type: array items: $ref: '#/components/schemas/AgentPluginConfigDto' maxActiveRootRuns: type: number description: >- Policy-enforced maximum number of concurrently admitted root runs for this agent. minimum: 1 maximum: 32 maxChildRuns: type: number description: Maximum number of delegated child runs allowed by policy. minimum: 1 maximum: 128 maxConcurrentChildRuns: type: number description: >- Policy-enforced maximum number of delegated child runs that one parent may start concurrently. minimum: 1 maximum: 32 maxTicks: type: number description: Maximum number of ticks allowed by policy. minimum: 1 maximum: 1000 childFailureMode: type: string description: How the parent reacts when one delegated child fails. enum: - fail_parent - merge_partial AgentSummaryDto: type: object properties: activeRunCount: type: number description: Number of active root runs. example: 1 queuedInboxCount: type: number description: Number of PENDING inbox items. example: 3 attachedPluginCount: type: number description: Number of attached registry or inline plugin references. example: 2 totalRunCount: type: number description: Total run count for this agent. example: 14 failedRunCount: type: number description: ERROR run count for this agent. example: 2 blockedRunCount: type: number description: ACTION_REQUIRED run count for this agent. example: 1 tokenUsage: description: >- Token usage aggregated from provider-bearing assistant messages linked to this agent. example: today: 1800 currentBillingPeriod: 28000 previousBillingPeriod: 24000 allTime: 84000 input: 61000 output: 17000 cachedInput: 12000 reasoning: 6000 threadCount: 12 allOf: - $ref: '#/components/schemas/AgentTokenUsageDto' tokenUsageByModel: description: >- All-time provider usage grouped by the model persisted on each assistant message. example: - model: configured-model input: 61000 uncachedInput: 49000 cachedInput: 12000 output: 17000 reasoning: 6000 billable: 67200 type: array items: $ref: '#/components/schemas/AgentTokenUsageByModelDto' tokenUsageSeries: description: Daily token usage for the most recent 30 UTC days. example: - period: '2026-07-13' input: 1200 output: 480 cachedInput: 300 reasoning: 120 toolTokens: 220 billable: 1530 type: array items: $ref: '#/components/schemas/AgentTokenUsagePointDto' taskActivitySeries: description: Daily root-run activity for the most recent 30 UTC days. example: - period: '2026-07-13' created: 0 queued: 1 running: 1 waiting: 0 success: 8 blocked: 1 failure: 2 cancelled: 1 total: 14 type: array items: $ref: '#/components/schemas/AgentTaskActivityPointDto' lastRun: description: Most recent run. example: agentRunId: aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa status: FINISHED allOf: - $ref: '#/components/schemas/AgentRunPublicDto' lastFailure: description: Most recent failed run. example: agentRunId: bbbbbbbb-bbbb-4bbb-8bbb-bbbbbbbbbbbb status: ERROR allOf: - $ref: '#/components/schemas/AgentRunPublicDto' lastBlocked: description: Most recent action-required run. example: agentRunId: cccccccc-cccc-4ccc-8ccc-cccccccccccc status: ACTION_REQUIRED allOf: - $ref: '#/components/schemas/AgentRunPublicDto' required: - activeRunCount - queuedInboxCount - attachedPluginCount - totalRunCount - failedRunCount - blockedRunCount - tokenUsage - tokenUsageByModel - tokenUsageSeries - taskActivitySeries AgentDefinitionDto: type: object properties: identity: type: string description: Stable operator-owned identity of the Agent. maxLength: 500 mission: type: string description: Stable operator-owned mission of the Agent. maxLength: 2000 responsibilities: description: Responsibilities owned by the Agent. maxItems: 24 type: array items: type: string operatingPrinciples: description: Operator-owned principles applied on every Agent turn. maxItems: 24 type: array items: type: string successCriteria: description: Persistent completion criteria for the Agent role. maxItems: 24 type: array items: type: string AgentMemoryPolicyDto: type: object properties: enabled: type: boolean description: >- Whether the runtime may load and persist durable cross-run memory for this agent. default: true maxRecallEntries: type: number description: >- Maximum number of memory records the runtime should recall into one run. minimum: 1 maximum: 64 maxRecallChars: type: number description: >- Total character budget reserved for compact recalled memory summaries across one run. minimum: 128 maximum: 64000 AgentReleaseConfigDto: type: object properties: versionLabel: type: string description: Operator-facing release label such as 2026.04.10 or canary-1. minLength: 1 changeSummary: type: string description: Short operator-facing change summary for the current definition. minLength: 1 AgentAutomationPolicyConfigDto: type: object properties: defaultDeliveryMode: type: string description: >- Default trigger delivery mode for automation when the caller does not override it explicitly. enum: - INBOX - DIRECT_RUN approvalRouting: description: >- Approval-routing metadata reserved for enterprise automation escalation and operator delegation. allOf: - $ref: '#/components/schemas/AgentAutomationApprovalRoutingDto' notificationPolicy: description: >- Notification and delivery policy reserved for idempotent inbox fan-out and future chat delivery surfaces. allOf: - $ref: '#/components/schemas/AgentAutomationNotificationPolicyDto' AgentPluginConfigDto: type: object properties: contractVersion: type: number description: Plugin contract version for the governed runtime surface. default: 1 minimum: 1 maximum: 1 id: type: string description: Stable plugin id used by the agent runtime. minLength: 1 kind: type: string description: Plugin execution provider kind. enum: - tools - mcp_server - skill - package default: tools name: type: string description: Operator-facing plugin name. minLength: 1 description: type: string description: Optional plugin description. routing: description: >- Canonical English capability metadata used for tool selection. Operator-facing content may remain localized. allOf: - $ref: '#/components/schemas/AgentPluginRoutingConfigDto' executionMode: type: string description: Execution mode expected for the plugin provider. enum: - sync - background componentId: type: string description: Component id executed when this plugin runs. minLength: 1 componentVersion: type: number description: Optional fixed component version. minimum: 0 componentToolset: description: >- Multiple governed tool contracts backed by the same component implementation. allOf: - $ref: '#/components/schemas/AgentPluginComponentToolsetDto' mcpServer: description: Remote MCP server configuration used when `kind` is `mcp_server`. allOf: - $ref: '#/components/schemas/AgentPluginMcpConfigDto' mcpToolset: description: >- Per-tool MCP governance and loading configuration used when `kind` is `mcp_server`. allOf: - $ref: '#/components/schemas/AgentPluginMcpToolsetDto' skill: description: Markdown skill configuration used when `kind` is `skill`. allOf: - $ref: '#/components/schemas/AgentPluginSkillConfigDto' package: description: >- Markdown package manifest and dependency ids used when `kind` is `package`. allOf: - $ref: '#/components/schemas/AgentPluginPackageConfigDto' inputSchema: type: object description: >- Strict structured input schema exposed to discovery and runtime validation. Must be an object schema with additionalProperties=false and descriptions on every declared property. outputSchema: type: object description: Structured output schema expected from the plugin provider. approvalPolicy: description: Approval contract enforced before execution. allOf: - $ref: '#/components/schemas/AgentPluginApprovalConfigDto' mutationIntent: type: string description: Declared mutation intent used for governance and approval safety. enum: - read - write - destructive - external reversibility: type: string description: Declared reversibility of the plugin side effects. enum: - reversible - operator_reversible - irreversible - unknown targetScope: type: string description: Primary runtime scope impacted when the plugin executes. enum: - assistant_thread - agent - tenant - external sideEffectSummary: type: string description: >- Concise side-effect declaration recorded in approval, evidence, and audit surfaces. sideEffectCategories: description: >- Normalized side-effect categories used for governance and operator review. type: array items: type: string timeoutMs: type: number description: Maximum runtime in milliseconds. minimum: 1000 example: 30000 resultMaxBytes: type: number description: Maximum serialized result size in bytes. minimum: 512 autoDiscovery: type: boolean description: >- Whether the plugin should be surfaced as summary-only discovery context before explicit attachment. default: false enabled: type: boolean description: Whether the plugin is enabled for execution. default: true AgentTokenUsageDto: type: object properties: today: type: number description: Billed tokens since UTC midnight. example: 1800 currentBillingPeriod: type: number description: Billed tokens in the current UTC calendar month. example: 28000 previousBillingPeriod: type: number description: Billed tokens in the previous UTC calendar month. example: 24000 allTime: type: number description: All billed tokens across linked threads. example: 84000 input: type: number description: Total input tokens across linked threads. example: 61000 output: type: number description: Total output tokens across linked threads. example: 17000 cachedInput: type: number description: Cached input tokens across linked threads. example: 12000 reasoning: type: number description: Reasoning tokens across linked threads. example: 6000 threadCount: type: number description: Number of distinct assistant threads included. example: 12 required: - today - currentBillingPeriod - previousBillingPeriod - allTime - input - output - cachedInput - reasoning - threadCount AgentTokenUsageByModelDto: type: object properties: model: type: string description: Execution model identifier used to group this usage. input: type: number description: Provider input tokens attributed to this model. example: 61000 uncachedInput: type: number description: Non-cached provider input tokens attributed to this model. example: 49000 cachedInput: type: number description: Cached provider input tokens attributed to this model. example: 12000 output: type: number description: Provider output tokens attributed to this model. example: 17000 reasoning: type: number description: Provider reasoning tokens attributed to this model. example: 6000 billable: type: number description: Billing-weighted provider tokens attributed to this model. example: 67200 required: - model - input - uncachedInput - cachedInput - output - reasoning - billable AgentTokenUsagePointDto: type: object properties: period: type: string description: UTC calendar day represented by this point. example: '2026-07-13' input: type: number description: Provider input tokens recorded on this day. example: 1200 output: type: number description: Provider output tokens recorded on this day. example: 480 cachedInput: type: number description: Cached input tokens included in input. example: 300 reasoning: type: number description: Reasoning tokens recorded on this day. example: 120 toolTokens: type: number description: Content tokens emitted by tool messages on this day. example: 220 billable: type: number description: Billing-weighted provider tokens for this day. example: 1530 required: - period - input - output - cachedInput - reasoning - toolTokens - billable AgentTaskActivityPointDto: type: object properties: period: type: string description: UTC calendar day represented by this point. example: '2026-07-13' created: type: number description: Root runs still in CREATED on this day. example: 1 queued: type: number description: >- Compatibility bucket for PENDING root runs that are not waiting on CHILDREN. example: 2 running: type: number description: Compatibility bucket for PROCESSING root runs. example: 1 waiting: type: number description: Compatibility bucket for PENDING root runs waiting on CHILDREN. example: 1 success: type: number description: Compatibility bucket for FINISHED root runs. example: 8 blocked: type: number description: Compatibility bucket for ACTION_REQUIRED root runs. example: 1 failure: type: number description: Compatibility bucket for ERROR root runs. example: 2 cancelled: type: number description: Root runs cancelled on this day. example: 1 total: type: number description: All root runs created on this day. example: 11 required: - period - created - queued - running - waiting - success - blocked - failure - cancelled - total AgentRunPublicDto: type: object properties: agentRunId: type: string description: Stable run id. example: aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa agentId: type: string description: Owning agent id. example: 99999999-9999-4999-8999-999999999999 agentSessionId: type: string description: Durable session shared by successor runs. example: abababab-abab-4bab-8bab-abababababab previousRunId: type: string description: Immediately preceding run in this session. example: aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa rootRunId: type: string description: Root run id for lineage. example: aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa parentRunId: type: string description: Parent run id for delegated child runs. example: bbbbbbbb-bbbb-4bbb-8bbb-bbbbbbbbbbbb status: type: string description: Current run status. enum: - CREATED - PENDING - PROCESSING - ACTION_REQUIRED - FINISHED - ERROR - CANCELLED example: PROCESSING waitReason: type: string description: Reason a run is waiting. enum: - NONE - CHILDREN - CAPACITY - SCHEDULE - RECONCILE example: CHILDREN triggerType: type: string description: Trigger type such as REQUEST, INBOX, SCHEDULE, EVENT, or RETRY. example: REQUEST triggerRefType: type: string description: External trigger reference type. example: scheduleId triggerRefId: type: string description: External trigger reference id. example: schedule-123 ownerId: type: string description: Owning user or group id. example: 22222222-2222-4222-8222-222222222222 ownerType: type: string description: Owner type, typically USER or GROUP. example: USER serviceAccountUserId: type: string description: Service account execution principal. example: 11111111-1111-4111-8111-111111111111 rootAssistantThreadId: type: string description: Root assistant thread for the run. example: cccccccc-cccc-4ccc-8ccc-cccccccccccc currentAssistantThreadId: type: string description: Current assistant thread for the run. example: dddddddd-dddd-4ddd-8ddd-dddddddddddd lastAssistantMessageId: type: string description: Last assistant message id. example: eeeeeeee-eeee-4eee-8eee-eeeeeeeeeeee nextWakeAt: format: date-time type: string description: Next scheduled wake timestamp. example: '2026-06-13T10:30:00.000Z' tickSeq: type: number description: Current tick sequence. example: 3 tickCount: type: number description: Number of ticks executed. example: 2 childCount: type: number description: Number of child runs created. example: 1 lastHeartbeatAt: format: date-time type: string description: Last heartbeat timestamp. example: '2026-06-13T10:20:00.000Z' checkpointSummary: type: string description: Bounded terminal or checkpoint summary. example: Run completed and wrote the release summary. state: type: object description: Run-scoped state for operator diagnostics. example: instruction: Review deployment result: type: object description: Terminal result payload. example: summary: Deployment looks healthy. error: type: object description: Terminal error payload. example: message: Run failed. createdAt: format: date-time type: string description: Creation timestamp. example: '2026-06-13T10:00:00.000Z' lastUpdateAt: format: date-time type: string description: Last update timestamp. example: '2026-06-13T10:15:00.000Z' version: type: number description: Optimistic locking version. example: 1 required: - agentRunId - agentId - agentSessionId - status - serviceAccountUserId - tickSeq - tickCount - childCount - state - createdAt - lastUpdateAt - version AgentAutomationApprovalRoutingDto: type: object properties: mode: type: string description: >- How non-interactive automation should behave when a later phase adds approval delegation. enum: - block - delegate - notify delegateUserIds: description: >- Users who should receive delegated approval or escalation when automation cannot continue autonomously. type: array items: type: string format: uuid notifyUserIds: description: >- Users who should be notified when automation reaches an approval gate or escalation path. type: array items: type: string format: uuid AgentAutomationNotificationPolicyDto: type: object properties: dedupeWindowSeconds: type: number description: >- Suggested dedupe window for downstream delivery or inbox fan-out keyed by one automation trigger. minimum: 0 maximum: 604800 maxAttempts: type: number description: >- Suggested maximum delivery attempts for downstream notification or future chat delivery workers. minimum: 1 maximum: 32 auditTrail: type: boolean description: >- Whether automation-triggered delivery should retain an audit trail inside durable trigger state and future delivery workers. default: true AgentPluginRoutingConfigDto: type: object properties: title: type: string description: >- Canonical English capability title used for tool selection. Display names may remain localized. summary: type: string description: >- Canonical English capability summary used for tool selection. Display descriptions may remain localized. AgentPluginComponentToolsetDto: type: object properties: defaultConfig: description: Default governance and loading policy for component-backed tools. allOf: - $ref: '#/components/schemas/AgentPluginComponentToolConfigDto' tools: type: object description: Component-backed tools keyed by stable capability name. AgentPluginMcpConfigDto: type: object properties: transport: type: string description: Remote MCP transport supported by the SaaS runtime. enum: - streamable_http - sse url: type: string description: Absolute HTTPS endpoint for the remote MCP server. manifestTtlMs: type: number description: Manifest cache time-to-live in milliseconds. minimum: 1000 connectTimeoutMs: type: number description: Connection and handshake timeout in milliseconds. minimum: 1000 example: 30000 toolNamePrefix: type: string description: Prefix used to derive local MCP tool names from remote tool names. minLength: 1 authMode: type: string description: Auth mode used when calling the remote MCP server. enum: - none - bearer - basic - header bearerSecretId: type: string description: Secret id containing the bearer token. minLength: 1 bearerSecretDataId: type: string description: Optional secret data revision id for the bearer token binding. minLength: 1 basicUsername: type: string description: Optional username for basic auth. basicPasswordSecretId: type: string description: Secret id containing the basic auth password. minLength: 1 basicPasswordSecretDataId: type: string description: >- Optional secret data revision id for the basic auth password binding. minLength: 1 headers: description: >- Additional header bindings resolved from tenant secrets before calling the MCP server. type: array items: $ref: '#/components/schemas/AgentPluginMcpHeaderBindingDto' AgentPluginMcpToolsetDto: type: object properties: defaultConfig: description: >- Default governance and loading policy for tools discovered from this MCP server. allOf: - $ref: '#/components/schemas/AgentPluginMcpToolConfigDto' configs: type: object description: Per-remote-tool overrides keyed by remote tool name. AgentPluginSkillConfigDto: type: object properties: format: type: string description: Skill content format. Markdown is the only supported v1 format. enum: - markdown default: markdown content: type: string description: >- Markdown instruction content loaded into the assistant runtime when the plugin is attached. format: markdown minLength: 1 pluginIds: description: Registry plugin ids declared as one-level skill dependencies. type: array items: type: string format: uuid maxPromptChars: type: number description: Maximum number of skill content characters allowed into one prompt. minimum: 256 maximum: 64000 AgentPluginPackageConfigDto: type: object properties: externalId: type: string description: Stable package identifier used by this plugin definition. minLength: 1 example: com.revo.notion format: type: string description: Package content format. Markdown is the only supported v1 format. enum: - markdown default: markdown content: type: string description: >- Markdown instruction content loaded into the assistant runtime when the package is attached. format: markdown minLength: 1 pluginIds: description: Registry plugin ids declared as package dependencies. type: array items: type: string format: uuid metadata: description: >- Disabled marketplace-facing package metadata reserved for catalog publishing. readOnly: true deprecated: true allOf: - $ref: '#/components/schemas/AgentPluginPackageMetadataDto' maxPromptChars: type: number description: >- Maximum number of package content characters allowed into one prompt. minimum: 256 maximum: 64000 required: - externalId AgentPluginApprovalConfigDto: type: object properties: mode: type: string description: >- How plugin execution approval is enforced before the provider call is made. enum: - none - always - conditional safetyTier: type: string description: Safety tier enforced for approval and mutation governance. enum: - read - write - destructive - external - execute AgentPluginComponentToolConfigDto: type: object properties: enabled: type: boolean description: >- Whether this component-backed tool is enabled for runtime selection and execution. deferLoading: type: boolean description: >- Whether this component-backed tool is deferred behind discovery instead of eagerly exposed. alias: type: string description: Explicit local tool alias exposed to the runtime when set. minLength: 1 title: type: string description: Operator-facing title for this component-backed tool. description: type: string description: Tool-specific description exposed to the assistant runtime. entrypoint: type: string description: >- Stable implementation entrypoint passed to the component in api.input().agentTool.entrypoint. Defaults to the tool key. minLength: 1 inputs: type: object description: >- Static runtime inputs passed to the component under api.input().agentTool.inputs. popResult: type: string description: >- Named component result to return when the component executor supports result popping. minLength: 1 inputSchema: type: object description: >- Strict structured input schema exposed for this tool. The component receives matching arguments under api.input().agentTool.args. Must be an object schema with additionalProperties=false and descriptions on every declared property. outputSchema: type: object description: Structured output schema expected from this tool. approvalPolicy: description: Per-tool approval contract enforced before execution. allOf: - $ref: '#/components/schemas/AgentPluginApprovalConfigDto' mutationIntent: type: string description: Declared mutation intent for this component-backed tool. enum: - read - write - destructive - external reversibility: type: string description: Declared reversibility for this component-backed tool. enum: - reversible - operator_reversible - irreversible - unknown targetScope: type: string description: Primary runtime scope impacted by this component-backed tool. enum: - assistant_thread - agent - tenant - external sideEffectSummary: type: string description: Concise side-effect declaration recorded for this tool. sideEffectCategories: description: >- Normalized side-effect categories used for governance and operator review. type: array items: type: string AgentPluginMcpHeaderBindingDto: type: object properties: name: type: string description: Header name sent to the remote MCP server. minLength: 1 secretId: type: string description: Secret id containing the header value. minLength: 1 secretDataId: type: string description: >- Optional secret data revision id when the binding should pin a specific secret revision. minLength: 1 value: type: string description: >- Optional static header value used when the binding does not read from a secret. prefix: type: string description: >- Optional static prefix added before the resolved secret value, for example `Bearer `. AgentPluginMcpToolConfigDto: type: object properties: enabled: type: boolean description: Whether the tool is enabled for runtime selection and execution. deferLoading: type: boolean description: >- Whether the tool is deferred behind discovery instead of eagerly exposed. alias: type: string description: Explicit local tool alias exposed to the runtime when set. minLength: 1 approvalPolicy: description: Per-tool approval contract enforced before execution. allOf: - $ref: '#/components/schemas/AgentPluginApprovalConfigDto' mutationIntent: type: string description: Declared mutation intent for this remote tool override. enum: - read - write - destructive - external reversibility: type: string description: Declared reversibility for this remote tool override. enum: - reversible - operator_reversible - irreversible - unknown targetScope: type: string description: Primary runtime scope impacted by this remote tool override. enum: - assistant_thread - agent - tenant - external sideEffectSummary: type: string description: >- Concise side-effect declaration recorded for this remote tool override. AgentPluginPackageMetadataDto: type: object properties: avatarUrl: type: string description: Package avatar or icon URL shown in package catalogs. format: uri websiteUrl: type: string description: Package website URL. format: uri documentationUrl: type: string description: Package documentation URL. format: uri termsOfServiceUrl: type: string description: Package terms of service URL. format: uri privacyPolicyUrl: type: string description: Package privacy policy URL. format: uri supportUrl: type: string description: Package support URL. format: uri publisherName: type: string description: Publisher display name. minLength: 1 publisherUrl: type: string description: Publisher website URL. format: uri license: type: string description: Package license label. minLength: 1 categories: description: Package catalog categories. type: array items: type: string tags: description: Package search tags. type: array items: type: string securitySchemes: bearer: scheme: bearer bearerFormat: JWT description: Bearer JWT token or API Key (sk-*). type: http ````