LayerMap
Call graph for coding agents: callers up to HTTP routes, callees, usages for TS/JS, Go, Python, Java
At a glance
Not tested yet- GitHub stars
- 10
- Weekly downloads
- 3,064
- Last commit
- 2026-11-19
- Licence
- Apache-2.0
- Transport
- stdio
- Auth
- none
npx -y layermap
What it does
LayerMap is an MCP server that gives coding agents a read-only map of a codebase. It answers what calls a function, what that function calls, and which HTTP routes, handlers, jobs and commands a change reaches. It maps TypeScript, JavaScript, Go, Python and Java using each language's own compiler or type checker. It is for developers using MCP clients such as Claude Code, Codex, CodeBuddy Code, Pi and DeepSeek Harness.

Tools it gives your agent
- Map callers: traces what calls a function up to 8 levels, to the HTTP routes, handlers, jobs and commands they start from.
- Map callees: traces what a function calls down to 8 levels.
- Code lookup: shows modules, each file's declarations, and every usage of a symbol.
project_check_changes: lists the routes, jobs and commands the uncommitted changes reach, what the map cannot see, and the existing tests related to them.
Quick start
- Install the plugin for your client. For Claude Code:
/plugin marketplace add coffeecoproject/layermap /plugin install layermap@layermap
- For Codex:
codex plugin marketplace add coffeecoproject/layermap codex plugin add layermap@layermap
- For CodeBuddy Code:
codebuddy plugin marketplace add coffeecoproject/layermap codebuddy plugin install layermap@layermap
- For Pi:
pi install npm:layermap
- For DeepSeek Harness:
npx layermap setup dsh
- Start a new session in a Git repository and ask as usual.
- In Claude Code, choose "don't ask again" when asked before a map tool runs, or run:
npx layermap allow claude
Add it to Claude Code from a terminal:
claude mcp add layermap -- npx -y layermap
Or add this to .mcp.json in your project root (Claude Code) or .cursor/mcp.json (Cursor):
{
"mcpServers": {
"layermap": {
"command": "npx",
"args": [
"-y",
"layermap"
]
}
}
}Environment variables
LAYERMAP_STOP_CHECK: set to0to turn off the plugin's request that the agent run the change check before it finishes.
Example prompts
- What calls
Storage.MarkFeedAsRead, and which HTTP routes reach it? - What does this function call, down to the lowest level?
- Which endpoints, jobs and commands do my uncommitted changes affect, and which existing tests relate to them?
- Show me the declarations in this file and every usage of this symbol.
What to watch out for
- It maps only TypeScript, JavaScript, Go, Python and Java.
- It does not link a field or type to the code reading it, so the change check searches for references instead.
- Environment variables and configuration keys are linked by name only.
- It never writes or runs tests.
- The first time a session starts a version of LayerMap, npm installs that pinned package into LayerMap's cache.
- Claude Code asks once before each map tool runs in a project unless you allow it.
- The README reports benchmark results on small samples of tasks written by LayerMap's authors.
Not tested by tools4agents yet. Read the source repository before you install it.
Similar tools
Everything
The official test server. Exercises every part of the protocol, so it is the fastest way to check your setup works at all.
Filesystem
Gives an agent scoped read and write access to directories you name explicitly. The usual first MCP server people install.
Git
Read, search, and manipulate local Git repositories. Lets an agent inspect history and diffs without shelling out.
Sequential Thinking
Gives a model a scratchpad for multi step reasoning, letting it revise earlier steps instead of committing to a first answer.