Give your AI coding tools precise repository context in VS Code with CodeMesh

The short answer

CodeMesh has its own VS Code extension, the same one that installs into Cursor and Devin. In plain VS Code, it installs and keeps a structural understanding of your repository in sync, so whichever AI coding tool you use inside VS Code can call on it for targeted context instead of relying only on file search.

The panel inside VS Code

Opened from the activity bar. Repositories on one tab, the agent connection on another, and what CodeMesh has saved you on the first.

CodeMesh panelrunning
The editor panel, running. This is the extension's own code, embedded.

What does CodeMesh add to VS Code?

VS Code itself is an editor, not an AI coding agent. CodeMesh's VS Code extension is the same extension that powers CodeMesh in Cursor and Devin; inside plain VS Code, it installs, syncs, and maintains a structural understanding of your repository (functions, classes, imports, calls and dependencies). Whichever AI tool you run inside VS Code can then call on that structural context through MCP, instead of relying only on its own file search and reads.

Why repository discovery can consume tokens

When an AI coding tool doesn't already understand how a codebase fits together, it has to find out: searching for relevant files, opening candidates, reading through them, and tracing imports and call sites by hand. Every one of those steps costs tokens before it produces a useful suggestion or edit.

Discovery can consume tokens before useful reasoning begins.

How the integration works

flow
1Developer
2AI tool in VS Code
3CodeMesh
4Structural repository context
5AI tool's response

When your AI tool needs to understand how part of the repository is structured, it can call CodeMesh's MCP tools instead of searching the filesystem from scratch, and reason over the structural context CodeMesh returns.

On data handling
CodeMesh stores your repository's parsed structure and source text in a graph scoped to your organization. Your machine and your coding agent never hold a database credential: every query is scoped by the credential itself, executed read-only, and only the result crosses back.

Installing the CodeMesh extension

The CodeMesh extension isn't yet listed on the VS Code Marketplace, so it installs from a direct download instead.

  1. Download the CodeMesh .vsix build for your platform.
  2. In VS Code, open the Extensions panel, click the ⋯ menu, and choose Install from VSIX…, then pick the file you downloaded.
  3. Open the repository you want to work in. CodeMesh begins building a structural understanding of it automatically.

Installers are available directly: macOS (Apple Silicon), macOS (Intel), Windows, Linux.

Pairing CodeMesh with your AI coding tool

VS Code itself isn't an AI coding agent; it's the editor. CodeMesh's extension installs and keeps a structural understanding of your repository in sync, but for that context to reach an AI tool, that tool needs to be able to speak MCP.

If you use an MCP-compatible extension inside VS Code, clicking Connect in the CodeMesh extension can configure that extension's MCP settings for you, the same way it does for Cursor and Devin. If you use GitHub Copilot Chat or another AI extension, whether it can connect directly depends on that extension's own support for custom MCP servers; CodeMesh doesn't control that, and we don't claim guaranteed compatibility with every VS Code AI extension.

Does CodeMesh replace my AI coding tool?

No.

CodeMesh doesn't write code, plan changes, or run commands. Whatever AI coding tool you already use inside VS Code keeps doing that work. CodeMesh is a context layer that tool can call into, when it supports MCP, instead of rediscovering the repository from scratch.

FAQ

Yes. It's the same extension that supports Cursor and Devin: VS Code, Cursor and Devin are all VS Code-compatible editors, and the extension has no editor-specific branching.
It can, if Copilot Chat supports connecting to custom MCP servers in your version of VS Code. That depends on Copilot's own capabilities, not on CodeMesh; we don't guarantee compatibility with every VS Code AI extension.
No. CodeMesh doesn't edit code or replace any AI extension. It's a context layer that an MCP-compatible tool can call into.
For an AI tool inside VS Code to use CodeMesh's structural context, yes: MCP is the protocol used to reach CodeMesh's tools.
CodeMesh stores your repository's parsed structure and source text in a graph scoped to your organization. Your machine and your coding agent never hold a database credential: every query is scoped by the credential itself, executed read-only, and only the result crosses back.

Continue learning

What Is CodeMesh?MCPThe Benchmark

Install CodeMesh in VS Code

Sync a structural understanding of your repository, then connect whichever AI coding tool you use.

Start Saving TokensSee the Benchmark