API Documentation

Authentication

Every request requires your organization API key as a Bearer token. Create and manage keys in your RoboWrite dashboard under Settings → API keys. Each key is bound to a single organization; all reads and writes are scoped to it — there is no cross-tenant access. A missing or malformed header returns 401.

Base URL

The surface is path-versioned. Breaking changes ship under a new version prefix.

Clients & User-Agent

api.roboad.ai sits behind Cloudflare bot protection, so non-browser clients must send a normal, non-empty User-Agent header. Bare default agents such as Python's urllib are blocked at the edge: you get a 403 with an empty body (Cloudflare error 1010) before the request reaches the API — so there is no { detail } envelope to read. curl, httpx, and requests send an acceptable UA by default; if you build raw urllib requests, set one explicitly.

Conventions

Pagination

List endpoints return a uniform envelope: { items, total, limit, offset, has_more }. Control with ?limit= (1–200, default 50) and ?offset=; total is the full filtered count. A few endpoints return their own documented shape (e.g. keyword inventory, topic scores, brief sections).

Errors

Failures return the documented status code with a { "detail": "…" } body. Common codes: 401 (bad key), 404 (not found or not owned by your org), 409 (wrong state), 422 (invalid input — unknown body keys are rejected), 429 (rate limited), 503 (retry). The one exception with no JSON body is the Cloudflare 403 above.

Idempotency

Spend-sensitive write endpoints accept an Idempotency-Key header; a retry with the same key replays the original response (with Idempotency-Replayed: true) instead of repeating the operation. It is required on POST /content/items and optional on POST /brands/{brand_id}/topics, POST /content/generate, and POST /content/items/{item_id}/generate. Other writes (e.g. POST /briefs) do not dedupe — a dropped-response retry can create a duplicate. A same-key retry replays the original result, so changing the body under the same key does not start a new operation — use a fresh key when changing generation options.

Generate content from a topic

The fastest path to a generation. The body is exactly { "topic_id": "<uuid>" } — it does not accept direction or citation_mode (those live on the item path, POST /content/items/{item_id}/generate). Send an Idempotency-Key.

POST /content/generate
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json
Idempotency-Key: <opaque-unique-key>

{ "topic_id": "<uuid>" }

The call returns 202 Accepted with a job to poll:

HTTP/1.1 202 Accepted

{
  "job_id": "…",
  "workflow_id": "…",
  "status": "running"
}

Poll the job

Poll GET /jobs/{job_id} until the status stops polling. content_item_id and content_version_id arrive on the polled job (not on the 202), populated by the time the run reaches a terminal or review state.

GET /jobs/{job_id}

{
  "job_id": "…",
  "status": "awaiting_review",
  "content_item_id": "…",
  "content_version_id": "…",
  "error": null
}

Read the content

Fetch the item with a page of its versions. versions is a pagination envelope, not a bare array — read versions.items (newest first, ?limit= 1–200 default 50, ?offset=). For a single version use ?latest_only=true or ?version=N. Add ?include=scoring for quality scores and ?include=sources for the research sources cited in each version (combine with a comma). Both are opt-in and additive — without include they are null; guard on presence, not on null vs [].

GET /content/{content_item_id}?include=scoring,sources,compliance,compliance

{
  "id": "…",
  "title": "…",
  "status": "draft",
  "versions": {
    "items": [
      {
        "version_number": 1,
        "markdown_body": "# …",
        "word_count": 1240,
        "sources": [
          { "marker": 1, "source_name": "…", "source_url": "https://…" }
        ]
      }
    ],
    "total": 1, "limit": 50, "offset": 0,
    "has_more": false, "next_cursor": null
  }
}

Endpoint reference

The full v1 surface. Every collection endpoint is org-scoped, filterable where noted, and paginated.

Rate Limits

A single fixed window applies per organization — there are no per-plan request tiers. Exceeding it returns 429; back off and retry.

Not in v1 yet

These are deliberately out of v1 — design around them:

• Publishing to a connected CMS — publish from the dashboard for now. (This is why headless generation for a no-CMS brand lands in awaiting_review.)
• Completion webhooks — completion is poll-based today (GET /jobs/{job_id}); push webhooks are planned.
• Direct document file-download URLs — metadata and summaries are available; original-file signed URLs are not yet exposed.
• Pillar-strategy endpoints — a separate upcoming capability.

Changelog

Highlights. The complete dated record — including additive changes — is in the full API guide. /v1 response shapes are not frozen; check it before pinning a client.