Context7 prompt
Set up the Context7 MCP server in my Claude Code so it pulls real, current library documentation instead of writing code from memory. Safety first: check whether an MCP server named "context7" is already configured (run `claude mcp list`). If it already exists, tell me, change nothing, and skip straight to the CLAUDE.md step below. 1. Add the server - it's free, hosted, and needs no API key. Use the USER scope so it works in every project: claude mcp add --scope user --transport http context7 https://mcp.context7.com/mcp 2. Then append this rule to my global CLAUDE.md (~/.claude/CLAUDE.md - create the file if it doesn't exist). Read it first and don't duplicate or overwrite anything; only append: ## Context7 When working with an external library, framework, SDK, or API, use the Context7 MCP server to fetch its current documentation before writing or changing code - even for well-known libraries. Prefer it over relying on training data. When you're done, run `claude mcp list` to confirm context7 is connected, show me the CLAUDE.md section you added, and remind me to restart Claude Code so the new server loads.
Claude Code answers from training data with a cutoff date. The moment a library ships a new version, Claude is quietly working from an old one - reaching for APIs that were renamed, options that were removed, and patterns that don't exist anymore. The code looks right, then breaks, and you lose an hour debugging something that was never your fault.
Context7 is a free MCP server that fixes this at the source. When Claude needs a library, it pulls that library's real, current documentation on demand - the actual API as it exists today, not a memory of it - and writes against that.
Once it's wired in, the current docs are one paste away: tell Claude to use Context7 (or add the rule below so it does it automatically) and you get code that matches the version you're actually on - working the first time, instead of confident guesses from last year.