MCP Launchpad CLI
The mcpl CLI searches tools across several MCP servers and calls them from the terminal, with OAuth sign-in done by a human
- 206 ★ on the repository
- Last commit
Extended access
Why this level
- Starts the servers from `mcp.json` and calls any of their tools with the rights and tokens you gave those servers
- Stores OAuth tokens and keeps a background daemon with open connections
Install
- Works in
Manual install
uv tool install https://github.com/kenneth-liao/mcp-launchpad.gitNeeds `uv`. Check with `mcpl --help`.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Starts the servers from `mcp.json` and calls any of their tools with the rights and tokens you gave those servers
- Stores OAuth tokens and keeps a background daemon with open connections
This is third-party code. Review the repository files before installing.
What it does
MCP Launchpad installs as the `mcpl` utility and gives an agent that can run shell commands access to the tools of several MCP servers without registering them in the client. Configuration is read from `mcp.json` in the project, from `.claude/mcp.json` or from `~/.claude/mcp.json`. `mcpl list` shows servers and their tools, `mcpl search "<query>"` searches tools across all servers and returns five results by default, and `mcpl inspect <server> <tool>` prints the full schema (add `--example` for a sample call). A call looks like `mcpl call <server> <tool> '{"param": "value"}'`. A background daemon holds the connections, starts on the first call and restarts automatically, and `mcpl session status` shows its state. Environment variables are substituted into the config through `${VAR}` from `~/.claude/.env` and `./.env`. An agent cannot authorize OAuth servers because a browser is needed: a human signs in first with `mcpl auth login <server>`.
Who it is for. People who connected many MCP servers to an agent and want to find and call the right tool by command instead of keeping every server's schemas in the client.
Good fit when
- An agent runs in Claude Code, Cursor or Windsurf with bash access and you have five or more MCP servers
- You want to learn which server can search issues in a tracker without browsing each one by hand
- A script or agent calls an MCP tool from the shell with JSON arguments and parses the answer
- After a token change you want to verify every connection at once with `mcpl verify`
Not a fit when
- The agent cannot run shell commands: the utility works only through bash
- There are one or two MCP servers and the client handles them directly without an extra layer
- The servers use OAuth and nobody can sign in: an agent cannot complete browser authorization
- You need a registry or a shared team gateway: `mcpl` is a local single-user utility
Example request
Use mcpl to find a tool for creating a task in the tracker, show its schema and call it with the title "Check the release"Limitations
- The utility is meant for an agent with command-line access. Clients without bash will not fit.
- For OAuth servers a human must sign in beforehand with `mcpl auth login <server>` or `mcpl list --refresh`, and `mcpl auth status` shows the result.
- Installation needs the `uv` manager, and the Python version comes from the repository's `.python-version`.
- The agent instructions have to be copied by hand from the repository's `CLAUDE.md` into your own `CLAUDE.md` or `AGENTS.md`. The repository states no license.
How to disable. Remove the utility with `uv tool uninstall mcp-launchpad`. `mcpl session stop` stops the background daemon, while `mcp.json` and issued OAuth tokens stay on disk until you delete them.
FAQ
How do I install MCP Launchpad?
You need `uv`. Then `uv tool install https://github.com/kenneth-liao/mcp-launchpad.git` and check `mcpl --help`. Servers are described in `mcp.json` in the project or in `~/.claude/mcp.json`.
How do I find the right tool among many servers?
With `mcpl search "<query>"`: it searches all servers and returns five results, and the `--limit` flag raises that. `mcpl inspect <server> <tool>` prints the chosen tool's schema.
Why can an agent not sign in to an OAuth server?
Authorization needs a browser and the agent does not open one. A human runs `mcpl auth login <server>` first, then `mcpl auth status` shows that every server is authenticated.
What if a server does not appear in the list?
Restart the terminal and run `mcpl list --refresh`. `mcpl verify` tests the connections, and `mcpl session status` shows the daemon state.
Where do environment variables come from?
From `~/.claude/.env` and `./.env`. In `mcp.json` the reference is written as `${VAR}`, for example for `GITHUB_TOKEN`.
Discussion
No questions or reviews yet.
Related
CLI toolsA universal MCP CLI client from Apify: persistent sessions, OAuth, JSON output for scripts and shell access to any MCP server for agents
A CLI that turns any MCP server, OpenAPI spec or GraphQL endpoint into commands at runtime, with no codegen
A CLI for any MCP server: list and call tools, resources and prompts, run mock and proxy modes, guard access and manage agent configs
An MCP bridge: a single run_python tool runs code in a container and calls your MCP servers
A zero-dependency Python CLI: one catalog of MCP servers and skills for every agent, with only a compact index in context
A CLI to run LLM prompts and act as an MCP client from the terminal, with any MCP server and any model provider
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| MCP Launchpad (this card) | no | 206 | |
| mcp-cli | no | 1,282 | a lightweight CLI to find and call tools; Launchpad adds a background daemon and OAuth sign-in |
| MCPorter | no | 5,112 | a CLI and TypeScript runtime with typed clients; Launchpad is limited to search and calls from the terminal |
README in short
The README is short. It starts with installation through `uv` and a command reference, then describes the `mcp.json` format and the order in which configs are found, agent integration and a separate note on OAuth. The repository also holds a proposal document and a note on Anthropic's advanced tool use.