# Plannotator Plannotator is an open-source review surface for AI coding agents; it lets you annotate the agent's plan before execution and review the resulting diff in a PR-style interface, then send structured feedback back to the agent with one click. The plan view supports inline annotations, deletions, replacements, and revision history between iterations. The code view shows uncommitted agent changes side-by-side or unified, with line-level comments. The whole loop closes the gap between "squinting at terminal text" and a proper review workflow. It runs locally; plans never leave the machine, and shared links carry their data encrypted in the URL itself. It works across agent harnesses; [[Claude Code]], [[Codex CLI]], [[GitHub Copilot]], [[Gemini]], [[OpenCode]], and Pi are all supported, with editor integrations for VS Code, Obsidian, Bear, and GitHub PRs. It is dual-licensed under MIT or Apache 2.0 and written in TypeScript. It targets teams that want a human-in-the-loop checkpoint in agent-driven development without giving up on autonomy. ## References - https://plannotator.ai/ - https://github.com/backnotprop/plannotator ## Related - [[Claude Code Review]] - [[AI Coding Maturity Levels]]