mcp-cli
The npx @wong2/mcp-cli command runs MCP servers from a Claude Desktop config, lists their tools and calls them from the terminal.
- 447 ★ on the repository
- Last commit
- GPL-3.0
Extended access
Why this level
- Runs third-party MCP servers on your machine and can pass them environment variables through `--pass-env`.
- Stores OAuth tokens locally until `purge` runs.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/wong2-mcp-cliDetects 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.
- Runs third-party MCP servers on your machine and can pass them environment variables through `--pass-env`.
- Stores OAuth tokens locally until `purge` runs.
This is third-party code. Review the repository files before installing.
What it does
mcp-cli is a command-line inspector for Model Context Protocol servers. With no arguments it reads the Claude Desktop config, and `-c` takes any file in the same format. A server can be started from npm, a local script, a Streamable HTTP address through `--url` or SSE through `--sse`. Interactive mode lists a server's tools, resources and prompts and calls them. Non-interactive mode suits scripts: `call-tool`, `read-resource` and `get-prompt` with arguments in `--args`. OAuth works for SSE and Streamable HTTP servers, and `purge` deletes stored tokens.
Who it is for. Authors and users of MCP servers who need to check a server outside an agent and see which tools it actually exposes.
Good fit when
- A new MCP server is being written and each tool is checked before it meets an agent.
- A server from the Claude Desktop config misbehaves and a clean call without a model is needed.
- A CI job calls `call-tool` and compares the answer.
- A remote OAuth server is tested through the `--url` address.
Not a fit when
- You want a web debugging interface instead of a terminal.
- You need to manage server installation for agents.
- The server must run only inside the agent itself.
Example request
Connect to the filesystem server from the config, list its tools and call `list_files` with no arguments.Limitations
The README covers only running through `npx` and states no versions or supported protocol revisions. The last commit was in June 2026. Without `--pass-env` shell environment variables are not passed to the server. Running `npx @wong2/mcp-cli npx <package-name>` executes third-party code on your machine. License is GPL-3.0. There is no graphical interface: everything happens in the terminal. Checking OAuth servers needs a browser sign-in and stores tokens in the tool's local storage.
How to disable. Nothing is installed: the tool runs through `npx`. Delete OAuth tokens with `npx @wong2/mcp-cli purge`.
FAQ
Where does the server list come from?
Without arguments, from the Claude Desktop config. The `-c config.json` flag takes a file in the same format.
How do I connect to an already running server?
For Streamable HTTP use `--url http://localhost:8000/mcp`, for SSE `--sse http://localhost:8000/sse`.
Can I use it in scripts?
Yes. Non-interactive mode runs `call-tool`, `read-resource` or `get-prompt` with no prompts, and arguments go in `--args '{}'`.
How do I pass environment variables to a server?
Add `--pass-env`: variables from the current shell are passed to the server.
Where are OAuth tokens stored?
Locally. Delete them with `npx @wong2/mcp-cli purge`.
Discussion
No questions or reviews yet.
Related
CLI toolsThe official MCP server debugger: web UI, automation CLI and terminal UI in one package
MCP Interviewer checks an MCP server before an agent connects: provider limits, tool tests and a Markdown report.
sunpeak tests any MCP server in local replicas of the ChatGPT and Claude hosts and builds MCP Apps with React
kogiQA MCP gives an agent a browser that finds page elements without selectors, for debugging web apps and writing e2e tests
A community Playwright MCP server: browser automation, screenshots, device emulation and test code generation
MCP server: an agent sends GET, POST, PUT, DELETE and PATCH to your REST API and gets back status, headers and body
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| mcp-cli (this card) | no | 447 | |
| MCP Inspector | no | 11,248 | the official debugger with a web interface; mcp-cli works entirely in the terminal and reads the Claude Desktop config |
| mcp-cli | no | 1,282 | a light CLI by another author for discovering and calling tools; wong2 mcp-cli adds starting servers from npm, SSE, Streamable HTTP and OAuth |
README in short
The tool runs MCP servers from various sources and shows their tools, resources and prompts. OAuth is supported for SSE and Streamable HTTP.