locus mcp

Multi-provider MCP server management — configure, sync, and manage MCP servers across AI coding agents.

Manage MCP (Model Context Protocol) servers across AI coding agents. Locus maintains a canonical config at .locus/mcp.json and syncs it to provider-specific formats (Claude, Codex).

Usage

locus mcp <command> [options]

Commands

Command
Description

add <template>

Add a server from a built-in template

add-custom

Add a custom MCP server

remove <name>

Remove an MCP server

list

List configured servers

sync

Sync config to provider-specific formats

test <name>

Test an MCP server connection

status

Show config and provider sync status

enable <name>

Enable a server

disable <name>

Disable a server


Built-in Templates

Locus ships with pre-configured templates for popular MCP servers:

Template
Description
Required Environment

postgres

Query and manage PostgreSQL databases

POSTGRES_CONNECTION

filesystem

Read, write, and manage files locally

github

Interact with GitHub repositories, issues, and PRs

GITHUB_PERSONAL_ACCESS_TOKEN

fetch

Fetch and process content from URLs

memory

Persistent memory and knowledge graph storage


How It Works

  1. Servers are stored in .locus/mcp.json — the canonical config shared across providers

  2. locus mcp sync translates the canonical config into provider-specific formats (e.g., .claude/mcp.json for Claude)

  3. Each server can be scoped to specific providers via the config

  4. Health checks verify that configured servers are reachable and responding


Examples

Add a server from a built-in template

Add a custom MCP server

Manage servers

Sync to providers

Check status

Last updated