Skip to content

urmzd/dotfiles / dot_agents/skills/create-llms-txt

create-llms-txt

Unverified

Generate a spec-compliant llms.txt file for any project by analyzing its codebase, docs, and public URLs. Use when creating an llms.txt, making a project LLM-discoverable, or when the user mentions "llms.txt" in the context of generating or scaffolding one.

Create llms.txt

Generate a high-quality, spec-compliant llms.txt that gives LLMs maximum understanding of a project with minimum tokens.

Spec (llmstxt.org)

The file uses markdown with this exact section order:

  1. H1 (required) project name
  2. Blockquote (optional) one-paragraph summary with key context
  3. Prose (optional) additional detail, any markdown except headings
  4. H2 sections (optional) lists of [name](url): description links
  5. ## Optional (special) links that can be skipped for shorter context

No other heading levels allowed. Every list entry must be a markdown hyperlink. Descriptions after : are optional.

# Project Name

> One-paragraph summary covering what it is, who it's for, and key capabilities.

## Docs

- [README](url): Installation, usage, examples
- [API Reference](url): Endpoint documentation

## Optional

- [CONTRIBUTING](url): How to contribute

Procedure

1. Discover project identity

Read these files (skip missing ones):

  • README.md name, summary, features
  • AGENTS.md architecture, key directories
  • package.json / Cargo.toml / pyproject.toml / go.mod name, description, repo URL
  • CHANGELOG.md exists? link it
  • CONTRIBUTING.md exists? link in Optional
  • LICENSE exists? link in Optional

Determine the repo base URL from git remote or manifest. All links must be absolute URLs pointing to the main branch (e.g. https://github.com/{owner}/{repo}/blob/main/{path}).

2. Identify key files

Scan for files that help an LLM understand the project:

  • Entry points (main.go, src/lib.rs, src/index.ts, app.py, etc.)
  • Public API surface (exported modules, route definitions, CLI entry)
  • Config files that define behavior (Justfile, Makefile, docker-compose.yml, pyproject.toml)
  • Schema definitions (OpenAPI, GraphQL, DB migrations)
  • Skills (skills/*/SKILL.md)

Limit to 6-10 key files. Prefer files that explain what the project does over internal implementation.

3. Write the blockquote

One paragraph, 2-3 sentences max. Cover:

  • What the project is
  • Who it's for / what problem it solves
  • Key differentiating capabilities

Do NOT pad with generic filler. Every word should help an LLM decide whether this project is relevant to a query.

4. Organize sections

Use these H2 sections (omit empty ones):

SectionContents
## DocsREADME, AGENTS.md, API docs, guides
## API / Key FilesEntry points, public API, config, schemas
## SkillsAgent skills if skills/ directory exists
## OptionalCONTRIBUTING, LICENSE, changelog, secondary docs

5. Write link descriptions

Each : description should be 3-10 words explaining what the LLM will find there. Skip the description only if the link name is already self-explanatory.

Bad: - [README](url): The README file Good: - [README](url): Installation, usage, and examples

6. Validate

Before writing the file, check:

  • H1 is the first line, matches project name
  • Blockquote is present and concise
  • All URLs are absolute and resolve to real files
  • Sections appear in order: Docs, API/Key Files, domain sections, Optional last
  • No H3+ headings used (spec only allows H1 and H2)
  • ## Optional is last if present
  • Total file is under 50 lines (aim for density)
  • No duplicate links across sections
  • Every list item has a markdown hyperlink

Gotchas

  • No relative links. LLMs fetch llms.txt from a URL. Relative paths break. Always use absolute URLs.
  • Don't link directories. Link to the actual markdown/source file, not a directory path. skills/configure-ai/SKILL.md not skills/configure-ai/.
  • The ## Optional heading is semantic. It tells LLMs they can drop these links when context is tight. Put truly skippable content there; not your API docs.
  • Don't duplicate the README. The blockquote is a summary for LLMs, not a copy of your README's first paragraph. Write it fresh with an LLM audience in mind.
  • Keep it short. An llms.txt that's 200 lines defeats the purpose. Aim for 20-50 lines. If you need more, your project needs better docs, not a longer index.
  • Repo URL detection: check git remote get-url origin and normalize to HTTPS. Strip .git suffix. Use blob/main/ for file links.
rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 467kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 196kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 196kUpdated Claude CodeCodexVerifiedInstall