Using with LLMs

The Trakk Agency API ships machine-readable context files you can drop directly into AI assistants and coding tools. Instead of describing the API yourself, hand the LLM a file and it will understand the full API immediately.

---

Available files

File	Best for
AGENT.md	Full reference - endpoints, schemas, auth, enums, content schemes. Use when writing or reviewing integration code.
SKILL.md	Task recipes - fetching report content, batch syncing, auditing coverage. Use when executing a specific workflow.
openapi.yaml	OpenAPI 3.0 spec. Use with MCP servers or any tool that can import an API definition directly.
llms.txt	Index of all context resources. Useful for tools that support the llms.txt standard.

---

How to use

Chat (Claude, ChatGPT)

Paste the contents of AGENT.md or SKILL.md at the start of your conversation, or attach the file if the tool supports uploads. Start with SKILL.md for task-oriented help, AGENT.md when you need the full schema detail.

Cursor / GitHub Copilot

Add AGENT.md to your project or reference it directly in a prompt:

@AGENT.md

Write a TypeScript function that fetches the latest completed report for a given project ID.

MCP servers

Point your MCP server at the OpenAPI spec and it will generate tool definitions automatically:

Spec: /v1/llm/openapi.yaml
Auth: Bearer token in Authorization header

---

Tips

• Use SKILL.md for task recipes and AGENT.md for schema details - combining both gives the best results.
• Always tell the LLM which project ID you are working with upfront to avoid placeholder code.
• Remind the LLM to read content_scheme before accessing keyword content - the shape of content and html varies per report type and is easy to get wrong.