RoamiRoami

Playbooks

Opinionated recipes: what to call, in what order, with a dry run before anything changes.

Hygiene before publish

Audit, dry run, human OK, apply, re-audit. Never publish from the agent.

  1. 1
    Audit contentRun the content audit to build the fix list.
  2. 2
    Audit mediaCheck alt text, titles, and folders.
  3. 3
    Dry-run fixesSee every proposed change without applying it.
  4. 4
    ApplyApply only what you approved.
  5. 5
    Re-auditCheck the result and update the list.

Ship a landing page

List and get the guide, shape its blocks, link places and stories, then hygiene-check.

  1. 1
    Find the landing pageList guides and open the one you want.
  2. 2
    Edit blocksShape the content blocks.
  3. 3
    Link placesAttach places and stories.
  4. 4
    Related postsLink related posts.
  5. 5
    AuditRun the hygiene check.

Connect MCP

Wire Claude, Cursor, or Gemini with OAuth or an API key over SSE (or Streamable HTTP for Gemini).

  1. 1
    Pick the clientClaude and Cursor use SSE. Gemini uses Streamable HTTP at /mcp.
  2. 2
    Paste the connector URL or create a keyOAuth needs no key. Keys live under Developers in the app.
  3. 3
    Smoke-testAsk: List my Roami projects.

Call the API

Public blog reads and protected CMS writes. Know which path you need.

  1. 1
    Choose public or protectedReads are public. Writes need a key.
  2. 2
    Use a bearer keySend your key as a bearer token or x-api-key.
  3. 3
    Check the envelopeRead the response shape before you rely on it.

Build a public site

Hand another AI the OpenAPI document and a prompt. Public GET only, with no keys in the frontend.

  1. 1
    Copy the promptStart from the provided prompt in Developers.
  2. 2
    Import the OpenAPI documentGive it to your AI.
  3. 3
    GET /api/publicRead posts from the public endpoints.
  4. 4
    Smoke-testCheck the output before you build more.

Connect your agent and see the first draft

Free during beta. No card. Agents cannot publish.

Start freeAudit my blog