ring:generating-llms-txt

LLMs.txt & AI Documentation Generator

When to use

• Creating a new llms.txt for a repository
• Auditing an existing llms.txt for completeness
• Generating CLAUDE.md or AGENTS.md for AI coding agents
• Improving AI readability of a repository

Skip when

• Repository already has a complete, up-to-date llms.txt
• Task is code implementation with no documentation scope
• Repository is private/internal with no LLM discoverability need

Related

Complementary: ring:running-dev-cycle, ring:implementing-tasks

Generates llms.txt, CLAUDE.md, and AGENTS.md for repositories.

Step 1: Analyze Repository

1. Read README.md — project name, description, purpose
2. Read CONTRIBUTING.md — build, test, lint instructions (if exists)
3. Read Makefile / package.json / go.mod — build system, language, dependencies
4. Scan /docs/ — available documentation
5. Scan /api/ or OpenAPI specs — API surface
6. Read existing llms.txt / CLAUDE.md / AGENTS.md (if mode=audit)
7. Identify: language, architecture, test framework

Step 2: Generate llms.txt

Follow llmstxt.org specification exactly:

# {Project Name}

> {One-line description: language, what it does, license.}

{Optional: architecture, key concepts, domain terminology needed to work with this project.}

## Docs

- [{Doc title}]({url}): {Brief description}

## API Reference

- [{API name}]({url}): {What this covers}

## Code

- [{Key module}]({path}): {What this module does}

## Optional

- [{Secondary resource}]({url}): {Description}

Rules:

• One H1 (project name), required
• Blockquote summary — required, include language and license
• H2 sections only (no H3+)
• Links: [title](url): description format
• File in repo root: /llms.txt
• Target: fits in ~2K tokens

MUST include: name, architecture overview, key domain concepts, links to README/CONTRIBUTING/API docs/key modules.

MUST NOT include: internal-only docs, CI/CD details, issue tracker, full dependency lists, changelog.