Back to directory

withmcp

A command-line tool for toggling a system-wide collection of MCP servers on and off, globally or per project, so a rarely needed server can stay out of an agent's context by default and be enabled only when it is actually wanted.

withmcp

withmcp manages a developer's system-wide collection of MCP servers, letting individual servers or named groups of them be toggled on or off, either for one run or permanently, and either globally or scoped to a particular profile. It is meant for cases such as keeping a narrow-purpose server out of an agent's context by default and enabling it only for the task that needs it, switching between different toolsets for different kinds of work, or turning on an extra set of tools only in a particular environment.

It works as a thin launcher: it reads its own configuration to build the list of MCP servers that should be active for the current directory, translates that into the format the chosen agent harness expects, and then launches that harness; it does not communicate with the MCP servers itself, and authentication still happens inside the harness after launch. Configured servers already set up natively for the selected harness are left out of the generated list, and it currently supports the Claude, Codex and a project called pi as launch targets.

More in MCP servers

See all 734
  • Light Cloud

    light-cloud.com

    A hosting platform that deploys a static site, a container, or a managed Postgres database straight from a GitHub, GitLab, or Bitbucket push, with a preview URL for every branch and an MCP server for deploying from Claude Code.

  • ForensicDbg

    forensicdbg.com

    A post-mortem debugger for native Windows x86 and x64 crashes, with an MCP interface so AI tools can consume crash-dump analysis directly.

  • Karada.ai

    karada.ai

    A CI/CD platform that auto-compiles existing APIs into Go MCP servers, attaches composable middleware, and exposes them through one gateway usable from Cursor, Claude Code, and Windsurf.