OpenAPI & llms.txt
Two machine-readable surfaces: an OpenAPI 3.1 spec to generate a typed client, and llms.txt files to drop the whole API into a model’s context.
OpenAPI 3.1
The spec describes the REST surface with HTTP Bearer auth and typed request/response schemas (WebSocket messages are included as component schemas).
openapi-python-client generate --url https://docs.conyr.ai/openapi.yamlValidate it:
npx @redocly/cli lint https://docs.conyr.ai/openapi.yamlThe spec tracks the core REST surface and is expanding to cover every endpoint. For the newest intelligence endpoints, the API Reference is the canonical field-level source.
llms.txt
Plain-text context files following the llms.txt convention, for when you want a model to understand the API rather than call a generated client.
| File | Use it for |
|---|---|
/llms.txt | A compact index — drop into a system prompt or context window |
/llms-full.txt | The full expansion when the model needs every detail |
Building an agent? The MCP server is usually the better path than a generated client — the tools, schemas, and tier gating are handled for you, and results come back already structured and cited.