Skip to Content
Docs are being rebuilt — start at Introduction → How it works.
API referenceConnect Claude, Codex & MCP

Connect Claude, Codex & MCP

View as MarkdownOpen in ChatGPTOpen in Claude

These docs are built to be read by AI coding tools. Point Claude, Codex, Cursor, or any LLM at the resources below and it can write a correct TeamMate integration with minimal hand-holding.

Per-page: copy or open in an assistant

At the top of every API page is an actions bar:

  • Copy as Markdown — copies the raw Markdown of the page to your clipboard, ready to paste into any chat.
  • View as Markdown — the same page as plain .md. You can also just append .md to any docs URL (e.g. …/api-reference/agents.md).
  • Open in ChatGPT / Open in Claude — opens the assistant pre-loaded with a prompt that points at this page’s .md URL.

Site-wide: llms.txt

The whole documentation set is published in the llms.txt  format, so an AI tool can ingest it in one shot:

Paste either URL (or its contents) into your assistant and ask it to build against TeamMate.

Machine-readable API: OpenAPI

The API publishes a standard OpenAPI 3.1 spec — the best thing to hand an AI tool or API client:

Both are public (no key needed to read them). Import the spec into Postman, Insomnia, Scalar, or an SDK generator to get typed clients instantly.

Recipe: Claude / Claude Code

  1. Create an API key and export it: export TEAMMATE_API_KEY=sk_prod_….
  2. Give Claude the context — paste the llms.txt URL and the OpenAPI spec URL:
Read https://docs.taqat.ai/llms-full.txt and https://tmmate.ai/api/v1/openapi.json.
The TeamMate public API base URL is https://tmmate.ai/api/v1 and it authenticates with
"Authorization: Bearer $TEAMMATE_API_KEY". Write a script that creates an agent and a
knowledge base, then queries the knowledge base.
  1. Ask it to run the calls. Because the spec is OpenAPI, Claude (and Claude Code) can reason about every endpoint, parameter, and response shape.

Recipe: Codex / Cursor / other AI IDEs

Same idea — add the spec and docs as context:

  • Add https://tmmate.ai/api/v1/openapi.json as a docs/URL source (or download it into the repo).
  • Add https://docs.taqat.ai/llms-full.txt for prose explanations and examples.
  • Store the key as TEAMMATE_API_KEY in your environment, never in code.

Using the API over MCP

The public API is plain REST with an OpenAPI spec, so any OpenAPI-to-MCP bridge can expose it as tools to an MCP-capable client (Claude Desktop, Claude Code, Cursor). Point the bridge at:

  • Spec URL: https://tmmate.ai/api/v1/openapi.json
  • Auth header: Authorization: Bearer sk_prod_…

The bridge turns each endpoint into a callable tool, so the assistant can create agents or query knowledge bases directly.

Not to be confused with MCP servers inside TeamMate (Integrations → MCP servers), which let your agents call external MCP tools. That’s a separate feature from calling TeamMate’s own API over MCP.

Copy-paste starter prompt

You are integrating with the TeamMate public API.
- Base URL: https://tmmate.ai/api/v1
- Auth: header "Authorization: Bearer $TEAMMATE_API_KEY" (a workspace API key, sk_prod_…)
- Spec: https://tmmate.ai/api/v1/openapi.json
- Docs: https://docs.taqat.ai/llms-full.txt
- Responses are JSON wrapped in { "data": ... }; errors are { "error", "code", "details" }.
Task: <describe what you want to build>.
Search & queryTroubleshooting