llm-docs（LLM 文档）

LLM Documentation (llms.txt)

Many libraries provide LLM-optimized documentation at /llms.txt following the llms.txt standard.

Available Sources

When to Use llms.txt

Use llms.txt when:

• Starting work with a library - Get the architectural overview first
• Context7 lacks recent updates - llms.txt is always current
• You need official API reference links - llms.txt links to authoritative docs
• Understanding library structure - See what sections/features exist

llms.txt vs Context7

Recommended workflow: Start with llms.txt for overview, then use Context7 for specific implementation details.

Workflow

1. Fetch llms.txt for Overview

Use Firecrawl to fetch the llms.txt file:

firecrawl_scrape({
  url: "https://effect.website/llms.txt",
  formats: ["markdown"]
})

This returns a structured overview with:

• Project description
• Key documentation sections
• Links to important resources

2. Fetch Full Documentation (if available)

For deeper context, some libraries provide llms-full.txt:

firecrawl_scrape({
  url: "https://effect.website/llms-full.txt",
  formats: ["markdown"]
})

Note: Full versions can be large. Only fetch when you need comprehensive documentation.

3. Follow Up with Context7

After understanding the structure from llms.txt, query specific topics:

context7_resolve_library_id({ libraryName: "effect typescript" })
→ /effect-ts/effect

context7_get_library_docs({
  libraryId: "/effect-ts/effect",
  topic: "schema validation with branded types"
})

Quick Reference

Frontend & UI

firecrawl_scrape({ url: "https://ui.shadcn.com/llms.txt" })     # shadcn/ui
firecrawl_scrape({ url: "https://tanstack.com/llms.txt" })      # TanStack (Router, Query, Table)

TypeScript Ecosystem

firecrawl_scrape({ url: "https://effect.website/llms.txt" })    # Effect TypeScript
firecrawl_scrape({ url: "https://zod.dev/llms.txt" })           # Zod schema validation
firecrawl_scrape({ url: "https://trpc.io/llms.txt" })           # tRPC typesafe APIs

Web Frameworks

firecrawl_scrape({ url: "https://elysiajs.com/llms.txt" })      # Elysia (Bun framework)
firecrawl_scrape({ url: "https://hono.dev/llms.txt" })          # Hono (Web Standards)
firecrawl_scrape({ url: "https://bun.sh/llms.txt" })            # Bun runtime

Database & ORM

firecrawl_scrape({ url: "https://orm.drizzle.team/llms.txt" })  # Drizzle ORM
firecrawl_scrape({ url: "https://www.prisma.io/llms.txt" })     # Prisma ORM
firecrawl_scrape({ url: "https://supabase.com/llms.txt" })      # Supabase

Authentication

firecrawl_scrape({ url: "https://www.better-auth.com/llms.txt" }) # Better Auth
firecrawl_scrape({ url: "https://clerk.com/llms.txt" })           # Clerk

Mobile & React Native

firecrawl_scrape({ url: "https://docs.expo.dev/llms.txt" })     # Expo
firecrawl_scrape({ url: "https://reactnative.dev/llms.txt" })   # React Native

Infrastructure & Deployment

firecrawl_scrape({ url: "https://vercel.com/llms.txt" })        # Vercel
firecrawl_scrape({ url: "https://turbo.build/llms.txt" })       # Turborepo

Payments

firecrawl_scrape({ url: "https://stripe.com/llms.txt" })        # Stripe

Testing

firecrawl_scrape({ url: "https://vitest.dev/llms.txt" })        # Vitest

Best Practices

1. Check llms.txt first - Before diving into code, understand the library structure
2. Use the registry - Reference llm-docs-registry.yaml for known URLs
3. Prefer llms.txt over scraping random pages - It's curated for LLM consumption
4. Combine sources - llms.txt → Context7 → specific doc pages
5. Cache when appropriate - For repeated work, save llms.txt content locally