Plannotator
A free, open-source tool for reviewing and annotating an AI coding agent's plan or code before it executes, with PR-style diffs and version history, in your terminal or editor.
🔗 Visit Plannotator
Description
AI coding agents can plan out a multi-step change and start executing before you've really looked at what they intend to do. Plannotator adds a checkpoint: it shows you the agent's plan or code as something you can annotate, comment on, and approve or reject, instead of just watching it run.
Plannotator provides plan review with annotations, deletion marks, and comments, plus a PR-style code review interface with side-by-side or unified diffs and version history that tracks plan revisions over time. It runs locally with no telemetry by default, has a terminal interface via slash commands, and integrates with VS Code, Obsidian, Bear, and GitHub PRs. It supports multiple AI coding agents — Claude Code, OpenCode, Pi, Codex — and offers encrypted URL sharing for collaborative review. The core is open source (MIT/Apache 2.0) and free; a hosted 'Plannotator Workspaces' tier is in development on a waitlist.
💬 Our review
The short version: Plannotator fills a real gap — most AI coding agent interfaces don't give you a proper review step before execution — and being free, local, and telemetry-free by default makes it low-friction to add to an agentic workflow.
Cursor's built-in diff review and GitHub Copilot Workspace both offer some review-before-apply functionality, but they're tied to their own specific editor or platform; Plannotator's differentiator is working across multiple agents (Claude Code, OpenCode, Pi, Codex) rather than locking you into one vendor's tooling. Plain `git diff` review after the fact is the free fallback most people already use, but it doesn't give you the plan-level annotation, version history, or structured approval workflow Plannotator adds before code even gets written. If you're running multiple AI coding agents and want a consistent review layer across all of them rather than relying on each tool's own built-in (and often more limited) review UI, Plannotator's free core is worth adopting.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Cœur open source gratuit (MIT/Apache 2.0) ; Plannotator Workspaces hébergé payant en liste d'attente
Pros
Fonctionne avec plusieurs agents IA (Claude Code, OpenCode, Pi, Codex)
Local par défaut, sans télémétrie
Historique de versions des plans avec diffs
Partage par lien chiffré pour collaboration
Cons
Offre hébergée encore en liste d'attente
Nécessite d'apprendre une interface supplémentaire
Moins intégré nativement qu'un outil propre à un seul agent