One command to give any repo an AI agent workflow: Backlog.md task tracking + Superpowers skills + a live project dashboard — installed in seconds, cleanly removable.
This repository also configures its own agents. See what super-backlog tells them →
Skill Claude Code
Summarize current project state from Backlog.md (task counts, in-progress work, milestones) and point to the Project Dashboard. Use when the user asks for status, progress, or a project overview.
Skill Claude Code
Convert an approved design/implementation plan (from brainstorming/writing-plans) into reviewed Backlog.md tasks with acceptance criteria, milestones and dependencies. Use after a design is approved, when the user asks to decompose work into tasks, or before starting planned work in this project.
Skill Claude Code
Enforce the human review checkpoint before implementation starts. Use at session start on an existing task, right after spec-to-backlog created tasks, or when the user asks to implement a specific task: present the task, its pipeline phase and acceptance criteria, and wait for explicit approval before any code.