Everything you need to understand your codebase.
Codewalk is a codebase intelligence layer: it builds a persistent dependency graph, indexes code and docs, and exposes that intelligence through a Web UI, an MCP server, and a REST API.
Capabilities
Dependency Graph
Extracts imports across 15+ languages via tree-sitter and builds a complete dependency map with automatic module detection.
Blast Radius
BFS on the reversed dependency graph shows transitive impact before you change a file or module.
Semantic Search & RAG
ChromaDB vector search with parent-child chunking, distance-based filtering, and corrective RAG for grounded answers.
Code Review
Multi-stage review with OWASP security checks, blast-radius context, stack signals, and team guidelines — works with or without an index.
Voice Interface
Hands-free coding: mic → faster-whisper transcription → tool routing → spoken answer via edge-tts.
Architecture Health
Graph stats, betweenness centrality, PageRank, circular-dependency detection, and fix suggestions.
Docs & Guidelines
Index team docs (.md, .pdf, .txt) and coding guidelines, then ask questions with source citations.
Local-First & Private
Embeddings, graph, and vector store run locally. No data leaves your machine unless you enable cloud mode.
Code review powered by the intelligence layer
Codewalk's review engine is built on top of the codebase intelligence layer. It doesn't just lint — it understands your architecture, knows what files are risky, and reviews with full context.
How it works
Works with or without indexing
Review runs on any git repo — no analyze step needed. The intelligence layer enhances it when available.
Severity levels
| Capability | Typical tools | Codewalk |
|---|---|---|
| Architecture awareness | No dependency graph | DuckDB + igraph: PageRank, fan-in, cycles, bottlenecks |
| Blast radius | — | "This file has 23 callers — review with extra care" |
| Works without indexing | — | Just needs a git repo (graph enhances but isn't required) |
| Batched for large PRs | Dumps everything at once | 3-5 files per batch, sorted by risk, host LLM stays focused |
| Custom rubrics | Limited | Per-language + per-framework + team guidelines |
| Fix application | Suggests only | Accept/reject → apply atomically → verify with tests |
| Re-review | — | Hide rejected findings and verify fixes in a fresh review |
Run a review from the API
POST /review gathers the diff, runs static analysis, batches files by risk, reviews with the configured LLM, and returns a structured verdict with findings.
curl -X POST http://localhost:8000/review \
-H "Content-Type: application/json" \
-d '{
"staged": false,
"target_branch": "master"
}'Three ways to use it
Web UI
Visual exploration: knowledge graph, module browser, blast-radius viewer, reading order, execution flow, chat, review, and admin.
MCP Server
41 tools exposed through the Model Context Protocol for VS Code Copilot, Claude Code, Cursor, and OpenAI Codex CLI.
REST API
35+ FastAPI endpoints for analyze, chat, search, review, docs, voice, and health. Integrate into scripts or CI/CD.
MCP tool examples
Indexing on your server. Query everywhere.
Deploy the Codewalk backend to your own server and the index, the reviews and the spend stay on infrastructure you control. Repositories are connected by an HTTPS clone URL and re-indexed on demand; an App-based connection with push-triggered indexing is planned, not shipped.
- · Atomic index publish — a run either lands or does not
- · Version-fenced re-indexing, so a stale write is refused
- · A mismatch between index and embedding model is a conflict, not a silent miss
- · Streaming progress for every stage of a run