From documentation
Generate llms.txt (and optionally llms-full.txt) following the llms.txt standard for LLM-friendly project discovery
How this skill is triggered — by the user, by Claude, or both
Slash command
/documentation:generate-llms-txtThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Generate `llms.txt` and optionally `llms-full.txt` following the [llms.txt standard](https://llmstxt.org/) so that LLMs can
Generate llms.txt and optionally llms-full.txt following the llms.txt standard so that LLMs can
quickly understand the project.
The output directory is $ARGUMENTS if provided, otherwise default to the project root (.).
Check for existing files: If <output-dir>/llms.txt or <output-dir>/llms-full.txt already exist, read them in
full so any hand-written sections can be preserved and updated rather than replaced.
Gather project information (read in this order, each layer fills gaps left by the previous):
a. Build manifests — read package.json, build.gradle.kts, settings.gradle.kts, Cargo.toml, go.mod,
pyproject.toml, or whatever build/manifest file exists. Extract the project name, description, language, and
dependencies.
b. README — read README.md (or README, README.rst, README.txt) for the project overview, features, and
usage.
c. Existing documentation — glob docs/**/*.md, doc/**/*.md, and any other common doc directories. Read each
file to understand module descriptions and API summaries.
d. Source code — glob source files (src/**/*, lib/**/*, cmd/**/*, app/**/*, etc.) and read them to identify
the public API surface (exported functions, classes, interfaces, types). Only use source code to supplement where
existing documentation is lacking.
Build the llms.txt structure following this format:
# ProjectName
> One-line project description.
Brief 2–4 sentence overview of what the project does, who it is for, and what problem it solves.
## Docs
- [SectionOrModule](relative/path/to/doc.md): One-line description of what this covers
## API
- [SymbolName](relative/path/to/source.ext): One-line summary of what it does
## Optional
- [Contributing](CONTRIBUTING.md)
- [Changelog](CHANGELOG.md)
- [License](LICENSE)
Rules for this structure:
# ProjectName heading and > blockquote description are required.## Docs section lists documentation files with relative links and one-line descriptions.## API section lists key public symbols with relative links to their source files. Include only the most
important entry points — not every function.## Optional section lists supplementary files (CONTRIBUTING, CHANGELOG, LICENSE, etc.) only if they exist.Decide whether to generate llms-full.txt: Count the number of source files and estimate the expanded content
size. Apply these thresholds:
llms-full.txt.llms-full.txt and note in the report that the project is too large.Build llms-full.txt (if generating): Use the same structure as llms.txt, but replace every link with the
actual content inlined directly below the heading. For documentation files, include the full Markdown content. For API
symbols, include the signature and doc comment (not the full implementation).
Preserve hand-written content: Any section in an existing llms.txt or llms-full.txt that is not derivable from
the project (e.g., custom descriptions, usage notes, special instructions for LLMs) must be carried over verbatim.
Write the files: Write llms.txt (and llms-full.txt if applicable) to the output directory.
Report: List every file written or updated. If llms-full.txt was skipped, explain why (file count or estimated
size). Flag any areas where project documentation was too thin to produce a useful summary.
# for the project name, ## for sections, and Markdown list items for entries.llms-full.txt, use --- horizontal rules to separate inlined content blocks.llms.txt with just the project name and a
placeholder description, and tell the user.npx claudepluginhub mattbobambrose/mattbobambrose-claude-skills --plugin documentationGenerates llms.txt and llms-full.txt from git repo wikis for LLM-readable project docs per llms.txt spec. Resolves repo URL/branch first.
Generates a standards-compliant llms.txt file for any website by crawling the live site or reading the codebase directly. Helps make websites discoverable by AI agents and LLMs.
Optimizes project docs and READMEs for LLMs like Claude/Copilot via c7score scoring, llms.txt generation, question restructuring, and quality analysis.