Todo Inspector Pro

Your agent found five things. You closed the terminal. They are gone.
Todo Inspector Pro gives AI coding agents and your team one task list, kept in the repo, visible in the Unity Editor.
Your IDE already lists TODO comments. This adds the two things it does not have: structure on the comment, and a second channel the agent can write to without touching the Unity Editor API.
Structured TODOs
A comment carries an owner, priority, state, id and related files — not only a message:
// TODO(@claude, high, src:agent, state:open): missing null check before the Rigidbody access
Filter on any of it. Double-click to land on the line. Your existing // TODO-ann-high comments keep working, unchanged.
A task store an agent can write
Work that does not belong on a specific line — a question, an unfinished refactor, a decision it is not allowed to make — goes into a plain JSON file at Assets/.todoinspector/todos.json. Agents write files; they cannot call the Editor API. In-code TODOs and store records merge into one list in the window.
Both sides write that file, so a save from the Editor is a merge, not an overwrite. An agent closing a task while your window is open does not get undone.
One click of setup
Agent Setup writes the syntax, the tag table and the store path into CLAUDE.md, AGENTS.md and .cursorrules, between markers, leaving the rest of the file alone. After that your agent emits tasks the tool round-trips. No prompt engineering, no pasting the format every session.