> ## Documentation Index > Fetch the complete documentation index at: https://docs.origami.chat/llms.txt > Use this file to discover all available pages before exploring further. # Send a follow-up run > Admits a new run on an existing agent. Same agent, same workspace, same conversation context. Responds `202 Accepted` with the initial run object; agent work runs in the background. Use this to answer a `needs_input` question — the agent picks up from the prior conversation. ## OpenAPI ````yaml /openapi-v2.yaml post /agents/{id}/runs openapi: 3.1.0 info: title: Origami Agent API version: '2.0' description: > The Origami **v2 API** is the single canonical public surface, organized into five segments — **Projects** (tenancy), **Agents**, **Workspace**, **Outreach** (campaigns, sequences, steps), and **Account**. Drive AI workers from a prompt, read and write table data (rows, cells, upsert, enrichment runs), manage documents and workspaces, control the sequencer and campaigns, run recurring scheduled agents, and read account/credit state. **Objects are self-describing** (Stripe-style): every object in a response carries an `object` field naming its type (`"agent"`, `"run"`, `"table"`, `"campaign"`, ...), and every list endpoint returns the one list envelope `{ "object": "list", "items": [...], "nextCursor": string|null, "url": string }`. **v1 is deprecated** (still fully functional — no removal date in this release). New integrations should target v2 only. See the [v1 → v2 migration guide](./api-v1-to-v2-migration) for the route map; in short: | v1 | v2 | | --- | --- | | `GET /tables` | `GET /api/v2/tables` | | `GET /tables/:id/rows` | `GET /api/v2/tables/:tableId/rows` | | `POST /tables/:id/rows` (insert) | `POST /api/v2/tables/:tableId/rows/upsert` (upsert; no bare insert) | | `GET /batches/:id` | `GET /api/v2/enrichment-runs/:runId` (`GET /api/v2/batches/:batchId` remains a deprecated alias) | | `GET /credits` | `GET /api/v2/account/credits` | Errors use the canonical envelope `{ error, code, details?, handoff? }`; any 4xx the user can fix in-app carries a forwardable `handoff` link. ## Tenancy — parent-wide keys & the `x-origami-project` header Every API key is **parent-wide**: bound to the parent (agency) organization and able to act on the parent and any of its **projects** (child orgs). Send the **`x-origami-project: `** header on any v2 request to scope it to that project — all org-scoped resources (workspaces, tables, agents, outreach) then operate inside the project. Omit the header to act on the parent. Two exceptions: `/projects/*` manages the projects themselves from the parent (the header is ignored there, never rejected), and `/account` is always parent-scoped. The header fails closed: a malformed id → `400 VALIDATION_ERROR`; an unknown, cross-parent, or deleted project → `404 PROJECT_NOT_FOUND`. The plan gate (`canUseApi`) and request-rate limits stay keyed to the parent, and the concurrent-agent pool is shared across the whole parent. Credits spent inside a project draw on the parent's shared wallet, subject to the project's optional `monthlyCredits` budget cap. ## Transport — async by default `POST /agents` and `POST /agents/:id/runs` respond `202 Accepted` with the initial run object (`status: "running"`) as soon as the run is admitted. The actual agent work runs in the background. Callers poll `GET /agents/:id/runs/:runId` until `status !== "running"` to read the terminal state. Every endpoint returns plain `application/json`. There is no NDJSON, no heartbeats, no `tail -1`, no pre-stream vs in-stream error branches. Closing the HTTP socket has no effect on the run — there is no `abortOnDisconnect` knob. SDKs hide the polling — `await origami.agents.create({ prompt })` returns the final terminal `Run` object. Direct REST consumers poll `GET /agents/:id/runs/:runId` and honor the `Retry-After` header it returns while the run is still `running` (currently `15` seconds). The header is omitted on terminal responses — that absence is the signal to stop polling. Runs typically take 1–5 minutes; polling faster than the hint does not make a run finish sooner. ## Pagination — cursor only Every list endpoint takes the same parameter pair — `limit` (default 50, max 100; row queries allow up to 200) and `cursor` (opaque; pass back the previous page's `nextCursor`) — and returns the one envelope `{ object: "list", items, nextCursor, url }`. `nextCursor: null` marks the last page. There is no `page`/`pageSize` and no `offset`. Per-endpoint filters (e.g. `?search=`, `?status=`, `?enabled=`) compose with `cursor`/`limit`. ## Idempotency Any `POST` may send an **`Idempotency-Key`** header (1–255 printable ASCII chars). The first response for a key is recorded and replayed verbatim for retries with the same key + request for 24 hours; replayed responses carry the `Idempotency-Replay: true` header. Reusing a key with a *different* method/path/body — or racing a still-in-flight original — is `409 IDEMPOTENCY_ERROR`. Row upserts additionally dedup on the body `batchId` (domain-level idempotency, independent of transport retries). ## Rate limits Same layers as v1: **per IP** (300 req/min) and **per organization** (100 req/min). The real scarce resource for agent runs is the per-org concurrent-agent slot — see `CONCURRENT_LIMIT_EXCEEDED`. | Scope | Limit | |-------|-------| | Per client IP | **300 requests / minute** | | Per organization | **100 requests / minute** | | Concurrent runs (per organization) | **plan-tunable** (1 on starter, 3 on pro, 10 on scale, 20 on ultra, unlimited on enterprise) | Rate-limit headers (`X-RateLimit-Limit-{IP|Global}`, `X-RateLimit-Remaining-*`, `X-RateLimit-Reset-*`) and `Retry-After` are inherited from v1. ## Plan-tier policy Agent-loop knobs (`maxSteps`, `timeoutSeconds`) are **not** request parameters — they're internal config gated by the plan tier. The chat agent's intrinsic 20-minute `AGENT_TIMEOUT_MS` is the wall- clock ceiling. | Plan | Default model | Available models | internal maxSteps | | --- | --- | --- | --- | | `starter` | `origami-lite` | `origami-lite` | 30 | | `pro` | `origami-max` | `origami-lite`, `origami-max` | 50 | | `scale` | `origami-max` | `origami-lite`, `origami-max` | 75 | | `ultra` | `origami-max` | `origami-lite`, `origami-max` | 100 | | `enterprise` | `origami-max` | `origami-lite`, `origami-max` | 200 | ## Errors All error codes are `UPPERCASE_SNAKE_CASE` and live on `code`. The response body always has `error` (human-readable) and `code`; error-specific fields (e.g. `creditsRequired`, `topUpUrl`, `currentRun`, `activeSessions`) live alongside at the top level when they're stable parts of the contract. Non-`completed` terminal *run* states (`timed_out`, `errored`, `cancelled`, `needs_input`, `step_cap_hit`) are surfaced on the `Run` object's single `status` field — not as HTTP error codes. Callers branch on the run object. Notable codes beyond the per-endpoint ones: `IDEMPOTENCY_ERROR` (409 — `Idempotency-Key` reused with a different request, or the original is still in flight), `TOO_MANY_ROWS` (413 — a CSV upsert exceeds the 100-row per-request cap), `CAMPAIGN_NOT_FOUND` (404 — missing or cross-org campaign id), `SEQUENCE_NOT_FOUND` (404 — missing or cross-org sequence id), and `PROJECT_NOT_FOUND` (404 — the `x-origami-project` header names an unknown, cross-parent, or deleted project). ## Webhooks Sequencer events fan out via HTTPS POSTs you configure in **Settings → Developers → Webhooks**. See the [webhooks reference](/webhooks/overview) and the machine-readable [openapi-webhooks.yaml](/openapi-webhooks.yaml) spec for working receiver scaffolding. servers: - url: https://origami.chat/api/v2 description: Production security: - bearerAuth: [] tags: - name: Projects description: >- Projects (child orgs) under the parent org — list, get, create, update, delete. Managed from the parent; the `x-origami-project` header is ignored on these paths. - name: Agents description: Create, list, and archive AI workers (a.k.a. agents). - name: Runs description: Send prompts to an agent, poll status, and cancel. - name: Tables description: Read table state and lifetime credit cost. - name: Workspace description: Workspaces, tables, columns, rows, cells, enrichment runs, and documents. - name: Campaigns description: >- First-class outreach campaigns — list, get, read people + stats, create + edit (agentic), launch/pause/resume, delete. - name: Sequences description: >- Read per-recipient sequences (with steps inline), stop and delete them. Content edits go through the agentic campaign edit. - name: Scheduled agents description: Recurring agents (cron) — CRUD, enable/disable, trigger, run history. - name: Account description: Org account state — plan, capabilities, credit balance. paths: /agents/{id}/runs: parameters: - $ref: '#/components/parameters/AgentId' post: tags: - Runs summary: Send a follow-up run description: | Admits a new run on an existing agent. Same agent, same workspace, same conversation context. Responds `202 Accepted` with the initial run object; agent work runs in the background. Use this to answer a `needs_input` question — the agent picks up from the prior conversation. operationId: sendRun parameters: - $ref: '#/components/parameters/IdempotencyKey' requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/SendRunRequest' responses: '202': description: Run admitted; agent work is in the background. content: application/json: schema: type: object required: - run properties: run: $ref: '#/components/schemas/Run' '400': $ref: '#/components/responses/ValidationError' '402': $ref: '#/components/responses/InsufficientCredits' '404': $ref: '#/components/responses/AgentNotFound' '409': $ref: '#/components/responses/AgentBusy' '429': $ref: '#/components/responses/ConcurrentLimit' components: parameters: AgentId: in: path name: id required: true schema: type: string format: uuid IdempotencyKey: in: header name: Idempotency-Key required: false description: | Transport-level replay protection — any v2 `POST` may send it (1–255 printable ASCII characters; UUIDs recommended). The first response for a key is recorded and replayed verbatim for retries with the same key + method/path/body for 24 hours; replayed responses carry the `Idempotency-Replay: true` response header. Reusing a key with a *different* request — or retrying while the original is still in flight — returns `409 IDEMPOTENCY_ERROR`. 5xx responses are never recorded (the retry re-executes). schema: type: string minLength: 1 maxLength: 255 pattern: ^[\x21-\x7e]{1,255}$ schemas: SendRunRequest: type: object required: - prompt properties: prompt: type: string minLength: 1 maxLength: 10000 focusTableIds: type: array items: type: string format: uuid maxItems: 50 default: [] attachments: type: array maxItems: 20 default: [] items: $ref: '#/components/schemas/UploadedRunAttachment' description: See CreateAgentRequest.attachments. model: type: string enum: - origami-lite - origami-max Run: allOf: - $ref: '#/components/schemas/RunSummary' - type: object required: - workspaceId - request - response - todo properties: workspaceId: type: string format: uuid request: $ref: '#/components/schemas/RunRequest' response: nullable: true allOf: - $ref: '#/components/schemas/RunResponse' todo: $ref: '#/components/schemas/RunTodo' UploadedRunAttachment: oneOf: - type: object required: - kind - documentId properties: kind: type: string enum: - document documentId: type: string format: uuid - type: object required: - kind - tableId properties: kind: type: string enum: - table tableId: type: string format: uuid description: >- A document (injected as an attached file) or a table (merged into focusTableIds). RunSummary: type: object required: - object - id - agentId - status - prompt - model - steps - startedAt properties: object: type: string const: run id: type: string agentId: type: string format: uuid status: $ref: '#/components/schemas/RunStatus' prompt: type: string description: | The user prompt that drove this run, truncated to 200 characters with an ellipsis when longer. Suitable for a run-history list item; for the full prompt use Get Run. model: type: string enum: - origami-lite - origami-max description: | Public model id the run actually executed on. When the run has no assistant message yet (still admitting), falls back to the plan default. steps: $ref: '#/components/schemas/RunSteps' startedAt: type: string format: date-time completedAt: type: string format: date-time nullable: true RunRequest: type: object required: - prompt - model - focusTableIds properties: prompt: type: string model: type: string enum: - origami-lite - origami-max description: | Public model id — `origami-lite` or `origami-max`. Echoes the caller's request (after legacy-alias normalization) or the plan default when the request omitted `model`. Internal chat-agent ids (e.g. `origami-lobotomized`) are never surfaced here. focusTableIds: type: array items: type: string format: uuid description: Echoed verbatim from the request. RunResponse: type: object required: - text - actions - tables - transcript - transcriptTruncated description: | `null` while `status === "running"`. Once terminal, an object with the agent's cleaned prose, the structured workspace mutations it performed (`actions[]`), the full TableObjects for every touched table (`tables[]`), and (optionally) the full message transcript. `tables[]` is the resolved version of `actions[].tableId` — same shape as `GET /api/v2/tables/{id}`, fanned out server-side. Pass `?include=stats` on the Get Run call to attach economics on each embedded TableObject. properties: text: type: string nullable: true description: | Cleaned user-facing assistant text — internal markup (``, ``, ``, history-tool markers) is stripped server-side, identical to the stripping in `transcript` content. `null` when `status` is `errored` or `timed_out` (we suppress partial prose on failed runs). actions: type: array items: $ref: '#/components/schemas/RunAction' description: | Structured workspace-mutation actions the agent performed, in the order they fired. Projected from the persisted `data-*` parts the chat-agent's execute-code event listener writes onto each assistant message — the same audit trail the desktop UI renders. Empty when the run did no workspace mutations. tables: type: array items: $ref: '#/components/schemas/Table' description: | Full TableObjects for every table this run touched (derived from `actions[].tableId` plus any `focusTableIds` the caller passed). Same shape as a single `GET /api/v2/tables/{id}` response. Empty when the run touched no tables. transcriptTruncated: type: boolean description: | True when `transcript` was truncated to its hard cap. Always present (defaults to `false`) so consumers don't have to do a presence-check on every poll. transcript: type: array items: $ref: '#/components/schemas/TranscriptMessage' nullable: true description: | Full message history projected onto the surface a desktop user with **dev mode off** would see. Specifically: - Assistant text is stripped of internal markup (``, ``, ``, history-tool markers). - Tool calls are limited to the interactive widgets a non-dev user actually sees: `ask-questions`, `present-plan`, `contact-support`, `suggest-document-write`. The code the agent ran (`execute-code`), the VFS reads (`list-dir`, `read-file`, `grep`), and operational tools (`flag-session`) are dropped along with their results. - `reasoning` parts (dev-only in the UI) are dropped. - Empty assistant bubbles (text that was entirely internal markup and no public tool calls) are dropped. Opt-in only via `?include=transcript` on `GET /agents/{id}/runs/{runId}`. Default is `null`. RunTodo: type: object required: - pendingQuestions - nextActions properties: pendingQuestions: type: array items: $ref: '#/components/schemas/PendingQuestion' description: | Structured questions extracted from the agent's most recent `ask-questions` tool call, if any. When non-empty AND the chat_stream completed cleanly, `status` is promoted to `"needs_input"`. Answer by sending a follow-up run on the same agent (`POST /agents/{id}/runs`) with the user's answer as `prompt` — any free-text string is accepted; the agent parses it as natural language. nextActions: type: array items: $ref: '#/components/schemas/NextAction' description: | Structured next-action suggestions extracted from the agent's `...` block, if any. Surface the `label` to the user; act on `type` (when present) as a typed CTA the host can fire directly. ErrorEnvelope: type: object required: - error - code properties: error: type: string code: type: string details: type: object additionalProperties: true handoff: type: object description: | Present on 4xx errors the caller's user can resolve in-app — a forwardable link on the canonical app origin. required: - kind - url - label properties: kind: type: string url: type: string format: uri label: type: string RunStatus: type: string enum: - running - completed - needs_input - step_cap_hit - incomplete - cancelled - errored - timed_out description: | Single discriminator for the run's lifecycle. `needs_input` means the run completed cleanly but the agent's last assistant message included an `ask-questions` tool call; answer by sending a follow-up run on the same agent. `incomplete` means the model finished the step but the AI SDK could not parse a tool call it tried to emit (or it hit the output-token cap mid-tool-call), so the multi-step loop ended without the work being done — recoverable by sending a follow-up run on the same agent. Deploy restarts are auto-followed server-side, so callers never observe a `superseded` value. RunSteps: type: object required: - completed - max description: | Agent step progress. `completed` is the number of multi-step loop iterations the agent has actually executed; `max` is the plan-determined hard cap. Once `completed === max`, the run's `status` becomes `step_cap_hit`. properties: completed: type: integer minimum: 0 max: type: integer minimum: 1 RunAction: type: object required: - type - tableId description: | A single workspace-mutation event the agent performed. Discriminated by `type`; `tableId` is always present. Other fields are type-specific. v2 wire vocabulary speaks "leads", not "rows": the row-mutation action types are `leads_added`, `leads_deleted`, `leads_restored`. properties: type: type: string enum: - table_created - column_added - columns_updated - columns_deleted - leads_added - leads_deleted - leads_restored - columns_restored - table_restored tableId: type: string format: uuid tableName: type: string nullable: true description: Present on `table_created` and `table_restored`. columnId: type: string format: uuid nullable: true description: Present on `column_added`. columnName: type: string nullable: true description: Present on `column_added`. count: type: integer description: | Present on `columns_deleted`, `leads_deleted`, `leads_restored`, `columns_restored`. Number of items affected. leadCount: type: integer nullable: true description: Present on `leads_added`. Number of leads inserted. deletedAt: type: string format: date-time nullable: true description: Present on `columns_deleted` and `leads_deleted`. Table: type: object required: - object - id - workspaceId - name - leadCount - columns - credits - cells - url - createdAt - updatedAt properties: object: type: string const: table id: type: string format: uuid workspaceId: type: string format: uuid name: type: string leadCount: type: integer description: | Non-deleted row count. v2 wire vocabulary speaks "leads" instead of "rows"; the underlying DB column is still `rows`. columns: type: array items: $ref: '#/components/schemas/TableColumn' credits: $ref: '#/components/schemas/CreditsLifetime' description: | Lifetime credit cost across every cell. Equals the sum of every `columns[].credits.lifetimeUsed`. cells: $ref: '#/components/schemas/CellsLiveness' description: | Table-level liveness rollup — sums every `columns[].cells`. `running > 0` means at least one cell on the table is still being processed (e.g., enrichment kicked off by a recent run hasn't fully settled). The calling agent should treat this as "still working", not "no data found". url: type: string format: uri description: >- Deep link a human can open. For programmatic row reads, use `GET /api/v2/tables/{tableId}/rows`. createdAt: type: string format: date-time updatedAt: type: string format: date-time stats: $ref: '#/components/schemas/TableStats' description: Present only when caller passes `?include=stats`. TranscriptMessage: type: object properties: role: type: string enum: - user - assistant - tool content: type: string toolCalls: type: array items: type: object properties: id: type: string name: type: string input: {} toolCallId: type: string name: type: string result: type: string truncated: type: boolean PendingQuestion: type: object required: - type - question - suggestedAnswers - freeformOption properties: type: type: string enum: - single-choice - confirm description: | Question type the agent emitted. `single-choice` lists options the user is expected to pick from; `confirm` is a single approval prompt. question: type: string description: Natural-language question text to surface to the user. suggestedAnswers: type: array items: type: string description: | Answers the agent proposed, in the order it listed them. For `single-choice` these are the agent's options; for `confirm` it is the single-entry array `["Confirm"]`. This is NOT a hard menu — the caller may always reply with any free-text via the follow-up run (see `freeformOption`). The field is here so a host LLM can bias its follow-up `prompt` toward what the agent expected. freeformOption: type: string description: | A stable, agent-agnostic CTA string clients can render as a final "your own answer" button next to `suggestedAnswers`. Always the literal `"Or something else"` in v2.0. example: Or something else NextAction: type: object required: - label properties: label: type: string description: Natural-language label the agent suggested. type: type: string enum: - upload_csv - find_more_leads - export_csv - input_required - send_messages - schedule_play description: | Optional typed CTA the host can act on directly. When absent, treat as a plain text suggestion the user can ask the agent to do next. tableSlug: type: string description: | Optional table reference for table-scoped types (e.g. `export_csv`). TableColumn: type: object required: - object - id - name - type - kind - slug - autoTrigger - credits - cells properties: object: type: string const: column id: type: string format: uuid name: type: string type: type: string description: | Raw DB column type. Common values: `static` (user-entered), `code` (user-TS that runs per row, includes enrichments, scoring, and sequence columns), plus other internal-typed kinds for future extensibility. Treat as an open string. Prefer `kind` for classification — it splits `code` columns into `enrichment`, `score`, and `sequence`. kind: $ref: '#/components/schemas/ColumnKind' description: | Sequence-aware classification of this column. `sequence` columns draft outbound sequences — find them here rather than inspecting `type` (they share the `code` DB type with enrichments). Sequence columns are created via the agent or column code, never set through the upsert API. slug: type: string nullable: true description: Stable url-safe slug, or null when the column has none. autoTrigger: type: boolean description: | True when the column runs automatically on new leads. False for columns that only run when explicitly invoked (e.g. exports, manual-trigger code columns). credits: $ref: '#/components/schemas/CreditsLifetime' description: | Per-column historical cost. The table-level `credits.lifetimeUsed` is the sum of every column's `credits.lifetimeUsed` — the per-column breakdown is the primary surface for "which column is driving cost?". cells: $ref: '#/components/schemas/CellsLiveness' description: | Per-column liveness. `running > 0` means cells in this column are still being processed; the agent should not conclude "data not found" on empty cells until `running === 0`. stats: $ref: '#/components/schemas/ColumnStats' description: Present only when caller passes `?include=stats`. qualification: $ref: '#/components/schemas/ColumnQualification' description: | Present only when caller passes `?include=stats` AND this column has a relevance condition (i.e. acts as a quality filter). CreditsLifetime: type: object required: - lifetimeUsed properties: lifetimeUsed: type: integer description: | Sum of all settled cell-run charges, in credits. Stable and additive — once a row enrichment finishes it never moves backwards. Reservations (in-flight cell_runs) are not included. CellsLiveness: type: object required: - running - errored description: | Snapshot of cell-pipeline activity. The v2 run-finished signal is independent of the cell-pipeline-finished signal — when a run completes, downstream cells often still enrich in the background. Surfacing these counts lets the agent distinguish "still working on it" from "tried and failed", which is the difference between waiting and reporting a negative result. properties: running: type: integer minimum: 0 description: | Number of cells currently being processed (a `cell_run` is in `waiting`, `queued`, or `running` for that cell). `running > 0` means the agent should NOT report "no data found" — wait and re-read. errored: type: integer minimum: 0 description: | Number of cells that hit a settled failure. Includes `errored`, `out_of_credits`, `subscription_required`, and `connection_required`. The user-driven `stopped` state is intentionally excluded — it's a cancellation, not a failure to surface. TableStats: type: object required: - creditsPerLead - creditsPerQualifiedLead - findMoreEstimatedCredits - qualification - funnel - running - leadSources - hasUnknownTotalWithMore description: | Forward-looking table economics — the same numbers the desktop UI shows at the top of the table. Only present on responses that opted in via `?include=stats`. properties: creditsPerLead: type: number description: Estimated credits per sourced lead (pre-qualification). creditsPerQualifiedLead: type: number description: | Estimated credits per qualified lead, including the loss from dedup, exclusion, and filter cascade. findMoreEstimatedCredits: type: number description: Per-execution cost of the "Find More" code, in credits. qualification: $ref: '#/components/schemas/TableQualificationStats' nullable: true description: | Null when the table has no relevance conditions configured. funnel: $ref: '#/components/schemas/TableFunnelStats' running: $ref: '#/components/schemas/TableRunningStats' leadSources: type: array items: $ref: '#/components/schemas/TableLeadSourceStats' hasUnknownTotalWithMore: type: boolean description: | True when ≥1 active lead source has unknown TAM AND `has_more=true` (e.g. Twitter / LinkedIn post search). Consumers should render "X found (more available)" instead of a numeric TAM when set. ColumnKind: type: string enum: - input - enrichment - score - sequence description: | Sequence-aware API classification of a column: * `input` — user-entered (`static`) column; the only kind writable via `POST /tables/{tableId}/rows/upsert`. * `enrichment` — `code` column that runs per row to fetch/compute a value. * `score` — relevance / fit-score column. * `sequence` — `code` column that drafts an outbound sequence (`ctx.upsertSequence(...)` / `.draftMessage(...)`, or one that already has sequences). List its sequences via `GET /tables/{tableId}/sequences`. Created via the agent or column code only — never via the upsert API. ColumnStats: type: object required: - avgCreditsPerRun - callRate - totalRuns properties: avgCreditsPerRun: type: number description: Average credits charged when this column actually runs. callRate: type: number minimum: 0 maximum: 1 description: Fraction of leads that triggered this column (0-1). totalRuns: type: integer description: Lifetime count of cell runs feeding the average. ColumnQualification: type: object required: - pass - fail - unsure - total properties: pass: type: integer fail: type: integer unsure: type: integer total: type: integer TableQualificationStats: type: object required: - rate - fetchRate - qualifiedLeads - effectiveLeadsPerQualified - estimatedQualifiedLeads properties: rate: type: number nullable: true minimum: 0 maximum: 1 fetchRate: type: number nullable: true minimum: 0 maximum: 1 description: Filter-only qualification rate (excludes dedup / exclusion). qualifiedLeads: type: integer effectiveLeadsPerQualified: type: number description: Empirical leads-processed per qualified lead. estimatedQualifiedLeads: type: integer nullable: true TableFunnelStats: type: object required: - totalSourcedLeads - postStaticLeads - dedupedLeads - dedupRate - excludedLeads - excludedRate properties: totalSourcedLeads: type: integer postStaticLeads: type: integer description: Leads surviving dedup + exclusion + required static filters. dedupedLeads: type: integer dedupRate: type: number nullable: true minimum: 0 maximum: 1 excludedLeads: type: integer excludedRate: type: number nullable: true minimum: 0 maximum: 1 TableRunningStats: type: object required: - runningLeads - oldestActiveRunStartedAt properties: runningLeads: type: integer description: Leads with active (queued / running / waiting) cell runs. oldestActiveRunStartedAt: type: string format: date-time nullable: true TableLeadSourceStats: type: object required: - id - name - avgCreditsPerLead - totalCredits - totalLeads - source properties: id: type: string format: uuid name: type: string avgCreditsPerLead: type: number totalCredits: type: number totalLeads: type: integer source: type: string enum: - historical - estimate description: | `historical` when the average is derived from real cost_events; `estimate` when it was computed by static regex analysis of the lead source's TypeScript code. responses: ValidationError: description: | Bad request body or semantic validation failure. Route-level codes include `VALIDATION_ERROR`, `MODEL_NOT_AVAILABLE`, `UNKNOWN_WORKSPACE`, `UNKNOWN_TABLE`, and `WORKSPACE_TABLE_MISMATCH`. content: application/json: schema: allOf: - $ref: '#/components/schemas/ErrorEnvelope' - example: error: Invalid request body code: VALIDATION_ERROR details: issues: - path: prompt message: Required InsufficientCredits: description: | Org out of credits at admission time. `402` can also be `SUBSCRIPTION_REQUIRED` when the plan does not include API access; both shapes use the same `{ error, code }` envelope. content: application/json: schema: type: object required: - error - code properties: error: type: string code: type: string enum: - INSUFFICIENT_CREDITS - SUBSCRIPTION_REQUIRED creditsRequired: type: integer creditsAvailable: type: integer topUpUrl: type: string message: type: string hint: type: string example: error: Insufficient credits code: INSUFFICIENT_CREDITS creditsRequired: 5 creditsAvailable: 2 topUpUrl: https://origami.chat/settings/billing?source=api message: | Origami needs at least 5 credits to start any run; you have 2. Top up: https://origami.chat/settings/billing?source=api AgentNotFound: description: Agent not found (or belongs to another org) content: application/json: schema: allOf: - $ref: '#/components/schemas/ErrorEnvelope' - example: error: Agent not found code: AGENT_NOT_FOUND AgentBusy: description: A run is already in flight on this agent content: application/json: schema: type: object properties: error: type: string code: type: string enum: - AGENT_BUSY currentRun: type: object properties: id: type: string startedAt: type: string format: date-time ConcurrentLimit: description: Org-wide concurrent-agent cap reached (shared with the Origami UI). headers: Retry-After: schema: type: integer description: >- Seconds. Median wall-clock of the last 200 terminal runs in this org (fallback 60). content: application/json: schema: type: object properties: error: type: string code: type: string enum: - CONCURRENT_LIMIT_EXCEEDED limit: type: integer activeSessions: type: array items: type: object properties: agentId: type: string name: type: string startedAt: type: string format: date-time canUpgrade: type: boolean securitySchemes: bearerAuth: type: http scheme: bearer bearerFormat: og_live_... ````