Constitution Format

To create machine-friendly guardrails for LLM agents, define a concise constitution.md file at your project root. This file establishes token-efficient rules, patterns, and conventions that automated agents can easily enforce and reason about during development workflows.

A project constitution is a structured, token-efficient governance document that codifies best practices for AI agents. It defines categories like Stack, Patterns, Conventions, and Testing to provide clear rules and anti-patterns for automated development workflows.

A constitution.md file should include sections for Stack, Patterns, Anti-patterns, Conventions, Testing, and Security. These structured categories ensure LLMs and automated agents follow project-specific guardrails and coding standards efficiently.

Yes, you can use a constitution format as a living document to maintain AI governance while scaling projects. It provides token-efficient rules that help automated agents enforce established patterns and conventions consistently across growing codebases.

The best way to write token-efficient documentation for LLM workflows is creating a constitution.md file with concise, machine-friendly rules. This approach ensures automated agents can process and enforce project guardrails without consuming excessive context window tokens.