Plannotator

Key Features

• •Plan Review allows you to annotate an agent's proposed plan before execution by selecting text, marking deletions, adding inline comments, or writing replacements. This exports as structured feedback the agent understands, giving you visual control over scope and direction with version history and diff views between iterations.
• •Code Review provides a PR-style diff viewer for agent-written code before committing, with side-by-side or unified diff views, file tree navigation, and line-level annotations for code suggestions. It applies the same workflow as human PRs to agent output, enabling you to stage files and maintain code quality.
• •Local Execution ensures plans never leave your machine, running entirely locally for data security and privacy. This is combined with encrypted sharing via URLs where data lives in the link itself, allowing safe collaboration without external data exposure.
• •Version History tracks every plan revision automatically, saving each iteration with diffs between versions. This provides full auditability of changes, allowing you to see how plans evolve and revert if needed, with draft auto-save ensuring annotations survive server crashes.
• •Slash Commands like `/plannotator-annotate` let you annotate any markdown file, spec, folder, or URL and send feedback, while `/plannotator-review` launches a diff viewer for local changes or GitHub/GitLab PRs. These commands integrate seamlessly into your terminal workflow for quick access.
• •Editor Integrations with VS Code open plans in editor tabs with diff views and editor annotations, while Obsidian and Bear integrations auto-save plans to your vault with frontmatter and tags. This connects agent workflows to your existing knowledge bases and coding environments.
• •Automatic Hooking intercepts your agent's plan step automatically, presenting the review workspace without manual intervention. This zero-configuration approach ensures a seamless review process that fits naturally into your existing agent interactions.
• •Team Collaboration via Workspaces (hosted layer) offers shared plans, review history, integrations, and verification standards for teams, while the local plugin remains free and open source. This enables scalable collaboration across engineering teams.

Use Cases

• •Refining an AI agent's implementation plan for a new feature by marking unnecessary sections for deletion and adding clarifying comments. This results in a more focused and accurate plan that aligns with project requirements before any code is written.
• •Reviewing agent-generated code for a bug fix using the diff viewer to spot logical errors and annotate line-level suggestions. The outcome is cleaner, correct code that can be committed with confidence, reducing post-commit debugging.
• •Sharing an agent's plan with team members via encrypted URL for collaborative feedback without exposing sensitive data. This ensures alignment across the team and incorporates diverse perspectives before execution.
• •Auto-saving agent plans to an Obsidian vault with frontmatter and tags for documentation and knowledge management. This turns agent workflows into searchable, organized references for future projects.
• •Reviewing a GitHub pull request by URL within Plannotator to provide diff annotations and code suggestions in a familiar interface. This extends the tool's review capabilities to external code changes beyond local agent output.
• •Using slash commands to annotate a markdown specification or folder structure and send structured feedback directly to the agent. This streamlines the feedback loop for non-code artifacts in the development process.

Who is this for?

Plannotator is for developers, engineering teams, and technical leads who use AI coding agents like Claude Code, OpenCode, Pi, Codex, Copilot, Gemini, Kiro, Droid, and Amp. It targets those needing oversight and collaboration in agent-assisted development, particularly teams wanting to review plans and code before execution, integrate agent workflows into existing tools like VS Code and Obsidian, and maintain code quality through structured feedback loops.