toolmenu
Test harness that checks an MCP server's tool menu, calls and releases like a client would
- Pricing
- Open source
- Platforms
- CLI, API, Linux
- Runs
- Locally
- Version
- 0.28.1
- Listing updated
- today
Similar MCP servers
About toolmenu
toolmenu is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.
The README says toolmenu walks through a server the way an MCP client does: it reads the tool list, makes calls, inspects results and OAuth sign-in, and compares each release against the previous one. Checks are tied to the MCP spec, the tool's own schemas and documented client limits, and a finding comes with a rule id and a suggested fix.
The project positions itself as deterministic: no LLM runs during checks, so the same server should produce the same findings on each run, which suits CI. It reports how many prompt tokens the menu costs, flags menus that change mid-conversation, and can post a single pull request comment with a pass or fail verdict.
It is aimed at MCP server authors and maintainers who want to catch breaking changes, schema problems and token bloat before clients or agents do. It does not judge whether an agent chooses the right tool.
Install
Run in a terminal
claude mcp add toolmenu -- npx -y toolmenuAdd to claude_desktop_config.json
{
"mcpServers": {
"toolmenu": {
"command": "npx",
"args": [
"-y",
"toolmenu"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"toolmenu": {
"command": "npx",
"args": [
"-y",
"toolmenu"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"toolmenu": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"toolmenu"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"toolmenu": {
"command": "npx",
"args": [
"-y",
"toolmenu"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"toolmenu": {
"command": "npx",
"args": [
"-y",
"toolmenu"
]
}
}
}Using several servers? Build one config file with the MCP config generator.
What it does
- Menu and schema checks
- Validates the tool list and results against the MCP spec and the tool's declared output schemas.
- Release diffing
- Compares two menus and classifies changes as breaking, minor or notice, with the semver bump needed.
- Scenario sessions
- Plays a scenario or auto-generates one, calling tools and listing the menu after each step.
- OAuth discovery check
- Walks an HTTP server's OAuth discovery as a spec-following client would, without logging in.
- Token cost estimates
- Estimates how many tokens each tool's name, description and schema add to the model's context.
- CI pull request comment
- Generates a GitHub workflow that posts one updating comment with a verdict on each pull request.
Good for
- Catch breaking changes to tool names or required parameters before a release ships
- Check that an MCP server's tool list is accepted by a strict client SDK
- Keep prompt caching effective by spotting menus that change between runs
- Gate pull requests on a deterministic MCP server review in CI
Pricing
Open source
MIT licensed, free to use.
toolmenu FAQ
Does it use an LLM?
No. The README says no LLM is involved, so the same server gives the same findings on every run.
How is it installed?
The README shows running it with npx toolmenu, and it needs Node 22 or later.
Does it test whether an agent picks the right tool?
No. The README says that needs evals with a model, which toolmenu does not run.
Which transports does it support?
The README says it starts stdio servers or connects to HTTP servers.
Is toolmenu free?
Yes. toolmenu is open source, so you can use and self-host it for free.
How do I add toolmenu to Claude Code?
Run this command in your terminal: claude mcp add toolmenu -- npx -y toolmenu
What are the best alternatives to toolmenu?
The closest alternatives to toolmenu are ToolFit + MCPCheck, Api Test MCP, Toolkien, Toolmesh and MCP Tool Poisoning Scanner.
Where this comes from
- Description: written from the project's README and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.