marcelo-earth/llms-txt-generator

Overview

This skill generates a standards-compliant llms.txt file for any project or website following the llmstxt.org specification. It produces a concise, LLM-friendly summary, prioritized links to docs and examples, and optional resources to help language models understand and use the project quickly.

How this skill works

The skill scans the project using file globs to find READMEs, docs folders, API references, examples, and entry points. It extracts project name, purpose, features, caveats, and key links, then formats them into the exact llms.txt structure (H1, short blockquote, categorized H2 sections). The output is written to the provided path or to llms.txt in the project root, and a short explanation of contents and rationale is appended.

When to use it

• You need a single, machine-readable summary so LLMs can understand your project quickly.
• Preparing project documentation for automated assistants or code-generation tools.
• Publishing a website or repo and wanting LLMs to index key docs and examples.
• Creating an AI-readable onboarding file before releasing a library or API.
• Converting scattered docs into a prioritized, concise resource list for models.

Best practices

• Keep the blockquote to 1–2 sentences focusing on purpose and critical context.
• Prioritize links: most important docs first, optional resources in an "Optional" section.
• Use clear, short descriptions for each link explaining what it covers.
• Include caveats and compatibility notes prominently if they affect usage.
• Prefer absolute URLs for websites and relative or raw GitHub URLs for repos.

Example use cases

• Generate llms.txt for an open-source library so chat assistants can answer usage questions.
• Create an AI-readable summary for a corporate API before exposing it to internal tools.
• Automatically add llms.txt during a release pipeline to keep model-facing docs current.
• Convert a docs site into a compact llms.txt so search LLMs index the most relevant pages.
• Produce a lightweight onboarding file for plugin ecosystems or app stores.