The Dewstack REST API returns your published documentation as JSON or markdown, so you can sync it into a search index, a support tool, or your own product's in-app help. Three endpoints, one API key, no SDK required.

Base URL:

https://app.dewstack.com/api/public/v1

Every request carries the key as a bearer token:

curl https://app.dewstack.com/api/public/v1/articles \
  -H "Authorization: Bearer dsk_3f9a1c47e0b28d5641fa9e7c30bd48a2159cd7e0"

Missing or malformed headers return 401 with the message "Missing or invalid API key. Pass Authorization: Bearer dsk_...."; an unknown or revoked key returns 401 Invalid or revoked API key.

List every published page

GET /v1/articles

Returns every published page in the key's workspace, up to 500, with the URL each one is served at. This is the endpoint to poll when you want to know what exists.

{
  "data": {
    "articles": [
      {
        "id": 41822,
        "name": "Point a custom domain at your docs",
        "slug": "custom-domain",
        "collection_slug": "hosting",
        "url": "https://docs.acme.com/docs/hosting/custom-domain",
        "updated_at": "2026-07-14T09:22:41.000Z"
      }
    ]
  }
}

updated_at is the field to diff against on a re-sync ā€“ fetch full content only for the pages whose timestamp moved.

Fetch one page

GET /v1/articles/{collection_slug}/{article_slug}
GET /v1/articles/{collection_slug}/{article_slug}?format=markdown

Without format, content is the page's HTML. With format=markdown, it is markdown ā€“ which is usually what you want if the destination is a language model, a chat tool, or anything that will re-render the content itself.

curl "https://app.dewstack.com/api/public/v1/articles/hosting/custom-domain?format=markdown" \
  -H "Authorization: Bearer $DEWSTACK_KEY"
{
  "data": {
    "article": {
      "id": 41822,
      "name": "Point a custom domain at your docs",
      "slug": "custom-domain",
      "collection_slug": "hosting",
      "url": "https://docs.acme.com/docs/hosting/custom-domain",
      "format": "markdown",
      "content": "## Add the CNAME record\n\nā€¦",
      "updated_at": "2026-07-14T09:22:41.000Z"
    }
  }
}

Ask a question

POST /v1/ask

Runs a question through SmartDocs ā€“ the same engine behind the Ask a question widget on your docs site ā€“ and returns a written answer grounded in your content.

curl -X POST https://app.dewstack.com/api/public/v1/ask \
  -H "Authorization: Bearer $DEWSTACK_KEY" \
  -H "Content-Type: application/json" \
  -d '{"question": "How do I move a collection into another tab?"}'
{
  "data": {
    "answer": "Open Settings ā†’ Customize ā†’ Spaces (Tabs)ā€¦",
    "conversation_id": "c8b1f2e4-ā€¦"
  }
}

Pass the returned conversation_id back on the next call to ask a follow-up in the same thread.

This endpoint spends message credits

/v1/ask draws on the same SmartDocs message quota as the widget on your docs site. When the quota is exhausted the endpoint returns 429. The other two endpoints are free, and so is MCP.

Errors

Status When
400 POST /v1/ask with no question
401 Missing, malformed, unknown or revoked key
404 No published page at that collection and page slug
429 Rate limit exceeded, or SmartDocs message quota exhausted
502 The SmartDocs service is unreachable

Limits

  • 60 requests per 60 seconds per key. Space out bulk syncs.
  • 500 articles maximum per response.
  • Published pages only. Drafts are never returned.
  • Reads are one workspace per key. To sync two workspaces, create two keys.
Configuring SmartDocs Enable it, train it, style the widget and embed it elsewhere. llms.txt and llms-full.txt Whole-corpus text files at your docs root.