context7（Context7 文档搜索）

Context7 - Codebase Q&A

Codebase-specific context provider that indexes documentation and answers questions about your codebase.

Status

✅ INSTALLED via npm

npm install -g @upstash/context7-mcp
which context7-mcp  # /home/opc/.nvm/versions/node/v22.20.0/bin/context7-mcp

Setup

Environment variables required:

export CONTEXT7_API_KEY="your-api-key"
export UPSTASH_REST_API_TOKEN="your-redis-token"
export UPSTASH_REST_API_URL="https://your-db.upstash.io"

Usage

CLI Commands

# Index a codebase
uv run {baseDir}/scripts/context7.py index /path/to/codebase

# Query the codebase
uv run {baseDir}/scripts/context7.py query "How does the memory system work?"

# List indexed codebases
uv run {baseDir}/scripts/context7.py list

# Clear index
uv run {baseDir}/scripts/context7.py clear

Unified Runner

# Using skill.py (preferred)
python scripts/skill.py context7 query "How does memory work?"

# Direct
uv run {baseDir}/scripts/context7.py query "How does memory work?"

Common Patterns

Pattern 1: Ask about a file

python scripts/skill.py context7 query "How does TOOLS.md work?"

Pattern 2: Understand a system

python scripts/skill.py context7 query "Explain the memory architecture"

Pattern 3: Find implementation

python scripts/skill.py context7 query "Where is the qmd search implemented?"

Pattern 4: Quick code lookup

python scripts/skill.py context7 query "What does the exec function do?"

Efficiency Tips

1. Natural language works - Ask full questions, not keywords
2. Be specific - "How does the memory consolidation work?" > "memory"
3. Use context - Reference files you want to understand
4. Chain queries - Ask follow-up questions for deeper understanding

When to Use

Why Context7?

Quick Reference

Files

• scripts/context7.py - CLI wrapper
• MCP server: context7-mcp

See Also

• memory-keeper skill - Persistent context across sessions
• Full research: memory/MCP-SERVERS-RESEARCH.md
• Context7 GitHub: https://github.com/upstash/context7