llms.txt¶
llms.txt is a machine-readable function catalogue placed at the project root, designed for LLMs with file access (GitHub Copilot, Cursor, etc.).
Purpose¶
Instead of requiring a running MCP server, llms.txt provides a static index of all shortfx functions with their signatures and descriptions. An LLM can read this file and perform text search to find the right function.
Format¶
# shortfx - Function Catalogue for LLMs
> shortfx: 2,932 functions across 6 modules.
## fxDate.date_operations
- add_time_to_date(date, amount, unit) → Add a specified amount of time to a date
- calculate_days_between_dates(start_date, end_date) → Calculate days between two dates
...
Each entry follows:
Usage by AI Agents¶
- The LLM reads
llms.txtfrom the project root - Searches for relevant function names by keyword
- Imports and calls the function directly in Python code
Keeping It Updated¶
llms.txt is updated whenever public functions are added, removed, or renamed. The file is generated from the registry.py introspection and should always match the current API surface.