Procm — Claude Code MCP server for process management
Procm lets an agent start and monitor project processes, such as a development server or a test watcher.
claude mcp add procm-mcp -- npx -y procm-mcpWhat Procm does
An agent can launch a process in a chosen working directory, inspect its status, and read its stdout or stderr. It can also restart or stop a process; processes it created are cleaned up when the MCP server terminates. This helps when an agent needs to inspect server or test output while tracking down a failure.
Process creation uses an allowlist. The agent can request permission for a command with allow-start-process; approval applies when the command and working directory match. Do not allow this tool to run without confirmation, since that would let the agent add arbitrary commands. To connect Procm to Claude Code, run claude mcp add procm -- npx -y procm-mcp.
What it helps with
- start a development server from the project
- check whether a process is running
- read a process's stdout or stderr
- restart or stop a process by ID
- allow a command for a specific working directory
Example prompts for the agent
- Start the development server in this project and inspect its output.
- Show stderr for process ID
123. - Restart process
123and check its status.
How to install Procm
Platform CLI
claude mcp add procm-mcp -- npx -y procm-mcpPlatform CLI
codex mcp add procm -- npx -y procm-mcpConfig file
{
"mcpServers": {
"procm": {
"command": "npx",
"args": [
"-y",
"procm-mcp"
]
}
}
}Config file
{
"servers": {
"procm": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"procm-mcp"
]
}
}
}Platform CLI
gemini mcp add procm npx -y procm-mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Procm
How do I install Procm in Claude Code?
Run claude mcp add procm -- npx -y procm-mcp. You can use claude mcp add procm-mcp -- npx -y procm-mcp to register the server under the other name.
What does Procm do?
It gives an agent tools to start and manage processes, inspect their status, and read stdout or stderr. Processes it created are cleaned up when the server exits.
How does Procm control which processes an agent can start?
A command must be added to the allowlist through allow-start-process. Do not permit that tool without confirmation, or the agent could add commands without approval.
