● Alpha Release • v0.1.0 - Features & API subject to change
→ Understand context instantly
Find relevant code and documentation instantly. Semantic search across markdown, TypeScript, Python. No more grep. No more manual browsing.
$ grep -r "searchTerm" .
$ // 500 results...
$ // 100 false positives
$ // Still confused
$ npx @ngao/search --setup
→ Indexed 1,200 files
→ Ready for search
// Get exactly what you need
Perfect results • In milliseconds
Find code in milliseconds. No waiting. No frustration. Just instant results.
Semantic indexing means fewer false positives. Find what matters, not everything.
Give Claude only relevant code. Lower API costs. Faster responses.
→ Refactoring
Locate every reference, dependency, and related code instantly. Refactor with zero risk.
→ Documentation
Search markdown documentation alongside source files. Everything in one place.
→ AI Development
Give Claude the right context. Smarter suggestions. Lower costs. Better code.
→ Onboarding
New to a codebase? Find patterns and structure instantly. Reduce learning time.
Set up ngao-search MCP in your project with a few simple steps.
Open your terminal and navigate to the project directory you want to search.
$ cd /path/to/your/project
# Run interactive setup
$ npx @ngao/search --setup
# This will:
→ Create .ngao-search.json config
→ Set up project isolation
→ Start MCP server automatically
Execute the setup command which will:
Once the server is running, ask Claude to read the ngao-search instructions so it understands all available capabilities:
# Just ask Claude:
"Read the ngao-search instructions"
# Claude now has access to:
# • Index your project
# • Search semantically
# • Get AI-optimized results
# • Track search history
# • Reindex on changes
# All automatically integrated
That's it! Claude now understands your codebase and can provide:
Semantic Search
Understands code meaning, not just keywords
Lightning Fast
Results in milliseconds with smart ranking
Persistent Storage
LanceDB integration persists data across restarts
AI-Optimized
Results formatted for LLMs with full context
to setup
token savings
faster search