| name | skill-mcp-manager |
| description | Lifecycle manager for building, configuring, and "turning on" Model Context Protocol (MCP) servers. - mcp - mcp_server - setup_mcp - turn_on_mcp - configure_mcp - "@gemini-conductor" - "@sre-expert" |
MCP Manager
Expert in managing Model Context Protocol (MCP) servers for the Expert Intelligence Hub.
Core Lifecycle
The MCP servers are managed via the lifecycle tool located at tools/mcp-servers/manage_mcps.sh.
Turning On All Servers
To "turn on" all servers (ensure they are downloaded, built, and configured for Gemini/VSCode):
./tools/mcp-servers/manage_mcps.sh ready
[!IMPORTANT]
All MCP repositories are localized within tools/mcp-servers/mcp-*/ and are ignored by Git. If you accidentally clone them to the workspace root, run the ready command again to ensure they are correctly placed and then delete the root folders.
Individual Commands
- Status:
./tools/mcp-servers/manage_mcps.sh status Show build and runtime status.
- Build:
./tools/mcp-servers/manage_mcps.sh build [server] Build specific or all servers.
- Config:
./tools/mcp-servers/manage_mcps.sh config Regenerate .gemini/settings.json and .vscode/mcp.json.
- Health:
./tools/mcp-servers/manage_mcps.sh health [server] Verify JSON-RPC capability.
Configuration Details
- Gemini CLI: Configuration is stored in
.gemini/settings.json.
- VS Code: Configuration is stored in
.vscode/mcp.json.
[!NOTE]
Ensure you have npx, uvx, and go installed for runtime or build-time dependencies of specific servers.
