Skip to main content

​
What to provide an AI coding tool

AI coding tools do not need to be trained on Argyle. Give them current Argyle context instead: start with llms.txt, use the OpenAPI specification for endpoint and schema details, and connect the MCP server when your tool supports MCP.

​
Resources

  • llms.txt — Use Argyle’s LLM-friendly docs index to find relevant guide, API reference, and OpenAPI pages.
  • OpenAPI specification — Use the consolidated Argyle API spec for generated clients, schema-aware API calls, and endpoint references.
  • Argyle Docs MCP server — Connect this in MCP-capable tools so the assistant can search and read Argyle docs directly.

​
Common ways to use these resources

  • In AI coding tools such as Claude Code, Codex, Cursor, and VS Code, add the MCP server if the tool supports remote MCP servers.
  • If the tool cannot connect to MCP or does not automatically discover Argyle docs, give it the llms.txt link first and ask it to follow the linked Markdown docs for the specific feature.
  • For API implementation work, include the OpenAPI specification so the tool can use current endpoint and schema definitions.

​
Safety

Keep API keys, secrets, customer data, and borrower data out of AI prompts unless your organization has approved that workflow.