Postify for AI agents

Everything on this docs surface exists in a machine-readable form. Point your agent here and it can learn the whole API without scraping HTML.

Machine-readable surfaces

llms.txt

Curated index of every docs surface, built for agents.

llms-full.txt

The complete docs corpus as one Markdown document.

Markdown twins

Append .md to any docs URL for its Markdown mirror — e.g. /docs/api/createPost.md.

OpenAPI 3.1 spec

All 17 operations with full request/response schemas.

The MCP server

Postify ships a remote MCP server — 10 tools over OAuth 2.1, so Claude, Cursor, and any MCP-compatible agent can draft, schedule, and analyze by chat. Publish and delete tools stay off until the org owner enables them.

Add Postify to Claude Code
$ claude mcp add --transport http postify https://app.usepostify.com/api/mcp

Endpoint: https://app.usepostify.com/api/mcp

Paste-prompt

Bootstrapping an agent to build against the REST API? Hand it this:

Agent paste-prompt

Help me build against the Postify API. Read https://usepostify.com/llms.txt first — it indexes Markdown twins of every docs page. Base URL: https://app.usepostify.com/v1, auth: Authorization: Bearer postify_live_… (I'll provide the key). Use Idempotency-Key on POST /v1/posts, honor Retry-After on 429, and note that delete/publish-now return 403 dangerous_ops_disabled until the workspace owner opts in.

Guidelines for agents

  • Auth with Authorization: Bearer postify_live_… — keys are minted by humans at Settings → API keys.
  • Honor Retry-After on 429 rate_limited. Do NOT retry-loop 429 quota_exhausted — that is the monthly allowance; check GET /v1/usage instead.
  • Always send an Idempotency-Key (a UUID) on POST /v1/posts so your retries never double-post.
  • Delete and publish-now need the workspace owner's dangerous-operations opt-in — expect 403 dangerous_ops_disabled otherwise. Prefer drafting and letting a human publish.
  • Dedup webhook deliveries on the webhook-id header — delivery is at-least-once.