Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@CodeWeaver Where do we validate OAuth tokens?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
We're really proud of CodeWeaver and think it's pretty awesome, but we can't maintain it anymore.
We're focused on something else.
CodeWeaver is (was?) a sophisticated, smart, code search tool with wide provider support. and it's still licensed under your choice of MIT or Apache-2.0.
So please, fork it and build something great!
Documentation • Installation • Features • Comparison
Related MCP server: PureContext MCP
CodeWeaver gives Claude and other AI agents precise context from your codebase. Not keyword grep. Not whole-file dumps. Actual structural understanding through hybrid semantic search.
CodeWeaver is Professional Context Infrastructure. With 100% Dependency Injection (DI) and a Pydantic-driven configuration system, it provides the reliability and extensibility required for industrial-grade AI deployments.
Example:
Without CodeWeaver:
Claude: "Let me search for 'auth'... here are 50 files mentioning authentication"
Result: Generic code, wrong context, wasted tokens
With CodeWeaver:
You: "Where do we validate OAuth tokens?"
Claude gets: The exact 3 functions across 2 files, with surrounding context
Result: Precise answers, focused context, 60-80% token reductionCodeWeaver is no longer in alpha!
Early Release (0.x): CodeWeaver is in active development. APIs may change between minor versions. It's very well-tested but still in 'it works on my machine' territory. Use it, break it, help shape it.
Feature | CodeWeaver | Legacy Search Tools |
Search Type | Hybrid (Semantic + AST + Keyword) | Keyword Only |
Context Quality | Exquisite / High-Precision | Noisy / Irrelevant |
Extensibility | DI-Driven (Zero-Code Provider Swap) | Hardcoded |
Reliability | Resilient (Automatic Local Fallback) | Fails on API Timeout |
Token Usage | Optimized (60–80% Reduction) | Wasted on Noise |
# Add CodeWeaver to your project
uv add code-weaver
# Initialize with a profile (recommended uses Voyage AI)
cw init --profile recommended
# Verify setup
cw doctor
# Start the background daemon
cw start📝 Note:
cw initsupports different Profiles:
recommended: High-precision search (Voyage AI + Qdrant)
quickstart: 100% local, private, and free (FastEmbed + Local Qdrant)Want full offline? See the Local-Only Guide.
🐳 Prefer Docker? See Docker setup guide →
Hybrid search (sparse + dense vectors)
AST-level understanding (27 languages)
Reciprocal Rank Fusion (RRF)
Language-aware chunking (166+ languages)
Automatic local fallback (FastEmbed)
Circuit breaker pattern for APIs
Works airgapped (no cloud required)
Pydantic-driven validation at boot-time
100% DI-driven architecture
17+ integrated providers
Custom provider API
Zero-code provider swapping
Live indexing with file watching
Diagnostic tool (cw doctor)
Multiple CLI aliases (cw / codeweaver)
Selectable profiles for easy setup
AI agents face too much irrelevant context, causing token waste, missed patterns, and hallucinations. CodeWeaver addresses this with one focused capability: structural + semantic code understanding that you control.
Curation over Collection: Give agents exactly what they need, nothing more.
Privacy-First: Your code stays local if you want it to.
Infrastructure over Tooling: Built to be the reliable foundation for your AI stack.
📖 Read the detailed rationale →
Official Documentation: docs.knitli.com/codeweaver/
Built with ❤️ by Knitli
This server cannot be deployed
Code intelligence for coding agents: semantic, AST, graph, and full-text search. 279+ languages.
Codebase intelligence for agents: 152 structured artifacts across 21 programs, one call.
Shared memory for coding agents. Stop re-explaining your codebase every session.
Codebase intelligence for AI agents — dead code, blast radius, ownership.