# Codewalk MCP server

> Enables AI agents to analyze codebases using dependency graphs, RAG, and tree-sitter parsing to provide architectural insights and automated code reviews.

- Repository: https://github.com/gupta29470/codewalk
- Page: https://hostedmcp.org/servers/gupta29470/codewalk
- Transport: —
- Stars: 5
- License: MIT

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `CODEWALK_PATH` | no | no |  |
| `CODEWALK_REPO_NAME` | no | no |  |
| `CODEWALK_REPO_TOKEN` | no | no |  |
| `CODEWALK_SERVER_URL` | no | no |  |
| `EMBEDDING_MODEL` | no | no |  |
| `LLM_MODEL` | no | no |  |
| `LLM_PROVIDER` | no | no |  |

## Tools (8)

- `codewalk_analyze_codebase`
- `codewalk_generate_config`
- `codewalk_find_circular_dependencies`
- `codewalk_get_architecture_health`
- `codewalk_get_review_details`
- `codewalk_accept_and_verify_fix`
- `codewalk_run_static_analysis`
- `codewalk_run_tests`

## FAQ

### How does Codewalk handle large codebases without high token costs?

It uses RAG to embed code into a vector database, retrieving only the relevant chunks for a query rather than sending the entire codebase to the LLM.

### Can I use Codewalk for automated PR reviews?

Yes, it features a context-enriched review pipeline that uses dependency graphs and blast radius analysis to identify risky changes and security issues.

### Does it support voice commands?

Yes, it includes a voice interface that allows you to talk to your codebase via microphone and hear answers read aloud.
