DevLens MCP
DevLens analyzes a repo once and serves agents ranked files, call flow and change impact over MCP, no file-by-file reading
- 70 ★ on the repository
- Last commit
Extended access
Why this level
- The server reads repository sources and sends code to the chosen model provider when summaries are on
- The internet installer runs a script on the machine and the web UI listens on a local port
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/devlensDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The server reads repository sources and sends code to the chosen model provider when summaries are on
- The internet installer runs a script on the machine and the web UI listens on a local port
This is third-party code. Review the repository files before installing.
What it does
DevLens analyzes a TypeScript, JavaScript, Python, Java, Go or Rust repository and stores a code graph with a search index. Each language has its own parser: the TypeScript compiler, Python `ast`, JavaParser, `go/types` and Rust `syn`. The graph therefore carries type-checked edges, framework routes for Next.js, FastAPI, Spring Boot, Gin and axum, and ORM reads and writes. Agents get five MCP tools. `resolve_context` takes a task description and returns a text packet within a token budget: an answer line, ranked nodes with a one-line meaning, call flow, files with matched-line snippets and an id map. `find_symbols` does BM25F name lookup, `get_node` returns full detail for a node with callers and callees, `impact` shows what breaks when a symbol changes or between two commits, and `repo` manages analyzed graphs. Per-node summaries (technical, business and security) come from a model, and the command asks for confirmation before spending tokens. Structure-only analysis runs with no model and no key. For people there is a CLI with a `--json` flag and a web UI with the graph, which runs from the source tree.
Who it is for. Developers of large TypeScript, Python, Java, Go or Rust repositories where the agent spends tokens hunting for the right files.
Good fit when
- Before an edit the agent must learn what breaks if a function in a service changes
- A question like how auth works turns into dozens of grep calls and one compact answer is needed
- Before a pull request review you need the change radius between two commits
- You want the repository as an interactive graph with callers and callees
- You need per-node security summaries with a severity rating
Not a fit when
- The project is written in PHP, C# or Ruby: there are no native parsers for them
- The repository is small and the agent copes with plain file search
- You need summaries with no external provider: that mode gives only the graph structure
- You need the web UI from the installed binary: it runs only from the source tree
Example request
Analyze this repository with DevLens and tell me which files and routes a change to `createOrder` touchesMCP
- Transport
- stdio, http
- Authentication
- not required
Limitations
- Native parsers cover five language families: TypeScript and JavaScript, Python, Java, Go and Rust. Analyzing Go, Rust, Java and Python needs their runtimes, and `devlens doctor` checks them.
- Per-node summaries need a model provider and a key, set up with `devlens init`. Without it only structure analysis remains.
- The graph is built once and ages as code changes: the `freshness` action of the `repo` tool checks it, and `devlens analyze` reruns the analysis.
- The license is AGPL v3: commercial use follows its terms.
How to disable. Remove the server from the agent with `claude mcp remove devlens` or by deleting the devlens block from the Cursor, Codex and Hermes config. The package comes off with `npm uninstall -g @devlensio/cli`, and saved graphs live in DevLens storage and are removed separately.
FAQ
How do I connect DevLens to Claude Code?
Run `npm install -g @devlensio/cli`, then `devlens analyze .` in the project root and `claude mcp add devlens -- devlens mcp`. The agent discovers the five tools through tools/list.
Does it need a model key?
No, for the graph structure alone: analysis runs offline. A key and provider are needed for summaries and `devlens init` sets them.
Which languages does DevLens parse?
TypeScript and JavaScript, Python, Java, Go and Rust, each with its own native parser. It understands React, Next.js, FastAPI, Django, Spring Boot, Gin, axum and more.
How do I see the graph visually?
Run the web UI with `bun start` from the source repo: a graph with callers, callees and security flags opens on port 3000. It is not bundled into the installed binary.
How does MCP over HTTP work?
`devlens mcp http -p 7000` starts the server on port 7000. By default it speaks over stdio, which is enough for Claude Code, Cursor and Codex.
Discussion
No questions or reviews yet.
Related
MCP serversA local code graph with version history: an MCP server that tells the agent who calls a function and what an edit breaks
An MCP server that turns a large codebase into a semantic feature graph for precise navigation and meaning-based search
An MCP server that indexes code with tree-sitter so an agent reads single symbols, not files, traces paths and weighs edits
An MCP server that builds a code graph once and serves the answer to the agent instead of re-reading files
Deep Graph MCP: a code graph of a repository for agents
Deep Graph MCP
An MCP server that gives an agent a repository code graph: search by meaning, symbol links and the places an edit would touch
Local code structure graph: builds a repository map with tree-sitter and serves the assistant only the context a change actually touches over MCP
README in short
The quickstart takes four steps: install the CLI, set up a model, analyze the repo and connect the agent. It continues with a table of five MCP tools, supported languages and frameworks, the web UI, CLI commands, a benchmark against other graph tools and provider setup.