Skip to main content

MCP Panel

The Model Context Protocol (MCP) allows AI agents to connect to external data and tools. Klip provides a dedicated management interface for configuring MCP servers across your agents.

MCP Panel​

To open the management panel:

  • Command: Klip: MCP: Open Management Panel (klip.openMcpPanel)
  • Dashboard: Click the MCP tab in the main Dashboard.

Managing Servers​

Adding a Server​

  1. Click Add Server.
  2. Name: Give it a unique ID (e.g., git, postgres).
  3. Command: The executable to run (e.g., npx, docker, python).
  4. Args: Arguments for the command (e.g., -y @modelcontextprotocol/server-git).
  5. Environment: Optional environment variables.

Using Presets​

Klip includes built-in presets for popular MCP servers:

  • Git: Access git repository history and diffs.
  • PostgreSQL: Query a database.
  • Filesystem: Access local files.
  • GitHub: Search issues and PRs.

To use a preset:

  1. Click Browse Presets (klip.openMcpPresets).
  2. Select a server and click Install.

Editing and Removing​

  • Click the Edit (pencil) icon to modify a server's configuration.
  • Click the Delete (trash) icon to remove it.

Configuration Scope​

MCP servers are configured per-agent, but Klip tries to synchronize them where possible.

  • Claude: Writes to ~/.claude/settings.json (or project config).
  • Gemini: Writes to ~/.gemini/mcp.json.

Backups​

Klip automatically backs up your MCP configuration before making changes.

  • Restore: Use Klip: MCP: Restore Backup (klip.restoreMcpBackup) to roll back to a previous state if something breaks.