mcpc CLI
A universal MCP CLI client from Apify: persistent sessions, OAuth, JSON output for scripts and shell access to any MCP server for agents
- 1,077 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Calls arbitrary tools on connected MCP servers
- Stores OAuth tokens and bearer keys
- Supports experimental x402 crypto payments
Install
- Works in
Manual install
npm install -g @apify/mcpcOr bun install -g @apify/mcpc.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Calls arbitrary tools on connected MCP servers
- Stores OAuth tokens and bearer keys
- Supports experimental x402 crypto payments
This is third-party code. Review the repository files before installing.
What it does
mcpc maps every MCP operation to a shell command: listing and calling tools, reading resources, prompts and async tasks. Connecting to a server creates a named session kept alive by a background process, so you can work with several servers at once. An agent with Bash access does not need dozens of tool definitions in context: it finds the right ones via --help and grep across sessions, and with --json writes scripts using jq and xargs. It supports OAuth 2.1 with credentials in the OS keychain, a proxy that keeps keys away from generated code, tool schema validation and experimental x402 payments.
Who it is for. For MCP server developers and anyone wiring MCP into CLI agents such as Claude Code and Codex.
Good fit when
- You need to inspect and debug an MCP server from the terminal
- The agent burns too much context on MCP tool definitions
- You want to automate a chain of MCP calls in a shell script or CI
Not a fit when
- The agent has no shell access
- Your client's built-in MCP support is enough and you use few servers
Example request
Use mcpc to connect to mcp.apify.com, find the actor search tool and look for a ready website scraperLimitations
Requires Node.js or Bun, or Homebrew. On headless Linux, credentials go to ~/.mcpc/credentials.json unless gnome-keyring is set up. The client itself uses no LLM or external services, but whether a given MCP server is reachable from Russia depends on its owner. x402 payments are experimental and run on Base.
How to disable. Close sessions with mcpc, delete saved auth profiles, then remove the package: npm uninstall -g @apify/mcpc or brew uninstall mcpc.
FAQ
Can I connect a local stdio server?
Yes, reference an entry from an existing config, for example mcpc connect ./.vscode/mcp.json:filesystem @fs.
How do I keep keys away from agent-written code?
A human authenticates, and mcpc exposes that session through a local MCP proxy. The agent and generated code connect to the proxy without seeing credentials.
Discussion
No questions or reviews yet.
Related
CLI toolsA 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 adapter that connects an MCP client limited to local stdio servers to a remote MCP server over HTTP or SSE, with authorization
A zero-dependency Python CLI: one catalog of MCP servers and skills for every agent, with only a compact index in context
A Python framework and CLI for MCP: write servers as plain functions, test tools and install servers into Claude Code, Cursor and other clients
A CLI to run LLM prompts and act as an MCP client from the terminal, with any MCP server and any model provider
README in short
The README explains the idea: a CLI as the agent's local interface to MCP, with MCP as the remote standard. It then covers install via Homebrew, npm or Bun, sessions, three auth approaches, a sandbox proxy, agent modes for tool calling and scripting, schema validation and x402 payments. The repository includes an mcpc skill for agents. Apache 2.0 licensed.