This is the human approval checkpoint that sits between AI agent planning and execution. It routes plan reviews, git diffs, and markdown specs into a browser-based annotation flow so you can mark up a proposal before committing to it. Works best with Claude Code, Gemini CLI, Codex CLI, and OpenCode when you already have a concrete artifact to review, not when you need to create the plan itself. The routing discipline is the real value here: it aggressively hands off plan creation to task-planning, broad PR policy to code-review, and UI critique to agentation, keeping its own job narrow and honest. If your agent produces implementation plans or diffs and you want a visual gate before merge or execution, this is the tool.
npx -y skills add akillness/oh-my-skills --skill plannotator --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills