llms-txt-support

npx skills add majiayu000/claude-skill-registry --skill llms-txt-support --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

# llms.txt Support Skill ## Purpose Single responsibility: Detect, fetch, and utilize llms.txt files that provide LLM-optimized documentation, enabling 10x faster documentation ingestion. (BP-4) ## Background The llms.txt standard (https://llmstxt.org/) provides a convention for websites to expose LLM-friendly documentation. Instead of scraping entire sites, check for llms.txt first. **File hierarchy (check in order):** 1. `llms-full.txt` - Complete documentation (largest) 2. `llms.txt` - Standard documentation 3. `llms-small.txt` - Condensed documentation (smallest) ## Grounding Checkpoint (Archetype 1 Mitigation) Before executing, VERIFY: - [ ] Base URL is accessible - [ ] Check all three llms.txt variants in order - [ ] Validate file content is actual documentation (not error page) - [ ] Confirm file size is reasonable for the documentation scope **DO NOT assume llms.txt exists. Always probe first.** ## Uncertainty Escalation (Archetype 2 Mitigation) ASK USER instead of guessing when: - Multiple llms.txt variants found - which size to use? - llms.txt content appears partial or outdated - File returns but content seems like error page - Site has llms.txt but content doesn't match

Check for llms.txt variants (in order of preference)
curl -I https://example.com/llms-full.txt
curl -I https://example.com/llms.txt
curl -I https://example.com/llms-small.txt
Check common alternate locations
curl -I https://example.com/.well-known/llms.txt
curl -I https://docs.example.com/llms.txt
Fetch and inspect first 100 lines
curl -s https://example.com/llms.txt | head -100
Check file size