Plannotator

Plannotator is a local, browser-based review surface for AI coding agents, covering Claude Code, Codex, Copilot CLI, Gemini CLI, OpenCode, Kiro, Droid, Amp and Pi. It hooks into each agent through its hooks, plugins, skills or slash commands. When the agent proposes a plan, writes HTML, or finishes writing code, the work opens in the browser, where you mark it up, comment, and send the feedback back to the agent to act on.

Plan review needs no command: it is wired into each harness's plan mode, so a markdown review surface opens whenever the agent creates a plan. Annotation also works on any markdown file, a folder of files, a fetched URL, an HTML file rendered as-is, or the agent's last message. Code review covers uncommitted local changes, GitHub pull requests, GitLab merge requests, GitButler workspaces, Jujutsu, Perforce and static diffs, with comments and code suggestions on the diff returned to the agent. Built-in AI can answer questions about what you are reviewing or run review agents that post comments to the diff.

Plans, diffs, annotations, drafts, history and configuration stay local, and Plannotator collects no usage telemetry. Content leaves the machine only for features that need it: URL annotation fetches the page, GitHub and GitLab review use the authenticated CLI, Ask AI sends context to the configured provider, and sharing sends the link. Small markdown shares are compressed into the URL fragment, while large markdown and raw HTML shares are encrypted with AES-256-GCM before upload to a self-hostable paste service that stores only ciphertext. Workspaces, a separate hosted product, is the main path for team sharing.

Herdr Annotate brings the same review flow to the terminal inside Herdr, annotating terminal text, markdown documents and agent replies, and a standalone terminal interface runs the document review without it. Annotations from terminal and browser reviews share one data store.