beads-superpowers

Process discipline and persistent memory for AI coding agents.

$ curl -fsSL https://raw.githubusercontent.com/DollarDill/beads-superpowers/main/install.sh | bash

View on GitHub

What is this?

AI coding agents have two recurring failure modes: no process discipline (they skip tests, rush to code, and claim work is done without verification) and no persistent memory (todo lists vanish when a session ends, the next session starts blind).

beads-superpowers is a Claude Code plugin that solves both at once. It merges Superpowers — a set of composable process skills enforcing TDD, brainstorming, systematic debugging, and two-stage code review — with Beads, a Dolt-backed issue tracker that gives agents persistent task memory across sessions.

The upstream Superpowers library ships 15 skills. This plugin extends it to 21, replaces all TodoWrite task tracking with bd beads commands, and wires everything together via a SessionStart hook that injects skills and task context automatically at the start of every session.

graph LR Hook["SessionStart Hook"] --> SP["using-superpowers"] Hook --> BD["bd prime"] SP --> Route{"Route to skill"} BD --> Route Route -->|design| B["brainstorming"] Route -->|plan| W["writing-plans"] Route -->|execute| S["subagent-driven-dev"] Route -->|debug| D["systematic-debugging"] Route -->|research| R["research-driven-dev"] S --> Review["Two-stage review"] Review --> Land["Land the Plane"] style Hook fill:#6366f1,color:#fff style Land fill:#f59e0b,color:#000

Documentation

Getting Started

Install, configure, and run your first project in 60 seconds.

Methodology

Design philosophy: why we merged Superpowers with Beads and how.

Skills Reference

All 21 skills — triggers, categories, and beads commands.

Example Workflow

Complete development lifecycle: 11-state FSM with diagrams.

Tips & Tricks

Command cheat sheet, troubleshooting, and best practices.

Skills at a Glance

Each skill is a composable process document that Claude Code loads on demand. The bootstrap skill (using-superpowers) routes to the right skill automatically based on the task.

brainstorming
Design before code writing-plans
Bite-sized task plans test-driven-development
RED-GREEN-REFACTOR systematic-debugging
4-phase root cause verification-before-completion
Evidence before claims requesting-code-review
Two-stage review receiving-code-review
Anti-sycophancy subagent-driven-development
Fresh agent per task executing-plans
Batch execution using-git-worktrees
Isolated branches finishing-a-development-branch
Merge + Land the Plane dispatching-parallel-agents
Independent tasks document-release
Post-ship doc audit getting-up-to-speed
Session orientation stress-test
Adversarial scrutiny project-init
Beads DB setup setup
Post-install config writing-skills
Meta: create skills auditing-upstream-drift
Staleness detection research-driven-development
Parallel research agents using-superpowers
Bootstrap + routing