Skip to main content
Glama

mcp/todo

Session checklist MCP server (stdio, newline JSON-RPC).

This is the TodoWrite / write_todos family: a session-scoped plan the model updates as it works. Not the MCP Tasks extension (tasks/list / long-running tools/call).

Tools

Tool

Description

todo_read

Return the current list (no args)

todo_write

Full-list replace (default) or merge: true by id

todo_write

{
  "todos": [
    { "content": "Fix auth redirect", "status": "in_progress", "activeForm": "Fixing auth redirect" },
    { "content": "Write tests", "status": "pending" }
  ]
}
  • Status: pending | in_progress | completed (aliases todo→pending, done→completed)

  • Required per item on replace: content (description / step / subject / title also accepted)

  • Optional: id (auto-slug if omitted), activeForm

  • merge: true patches by id (Cursor-style partial update)

  • plan / task_list accepted as aliases for todos

Ids are assigned on write (fix-auth-redirect, …). Exactly one in_progress is a convention in the tool description, not a hard error (a note is appended if more than one).

The list is in-memory (lives with the MCP process) unless MCP_TODO_PATH is set.

Related MCP server: todo-mcp-server

Env

  • MCP_TODO_PATH — optional JSON file { "todos": [...] } to persist across restarts

Run

From the monorepo root:

bun ./todo/server.mjs
# or persist:
MCP_TODO_PATH=/tmp/todos.json bun ./todo/server.mjs

Angela

Opt-in preset todo (prefix: false so the model sees todo_read / todo_write):

mcp: ['file-io', 'term-local', 'todo']

Related MCP Connectors

Related MCP Servers

  • F
    license
    Not graded
    quality
    D
    maintenance
    Enables AI clients to manage todo tasks through Tools for adding, listing, completing, and searching, alongside Prompts for daily reviews and task breakdowns. Exposes task lists and statistics as Resources with local JSON file persistence.
    -
  • F
    license
    Not graded
    quality
    D
    maintenance
    Enables Gemini CLI and other AI assistants to manage todos (create, read, update, delete, filter) through natural language commands via the Model Context Protocol.
    -
  • A
    license
    A
    quality
    D
    maintenance
    Enables AI assistants to manage a todo list with add, list, complete, and delete tasks.
    4
    6 npm
    MIT