obra/superpowers
A process-first skill framework for Claude Code: brainstorming, test-driven development, systematic debugging, and verification skills that gate how the agent works, not just what it knows. Agentiquette scores it 83/100 (A-tier), methodology v1.0, scored 2026-07-07.
Key facts
| Category | Framework |
|---|---|
| License | MIT |
| Maintainer | obra (Jesse Vincent) |
| Compatible agents | claude-code |
| File patterns | SKILL.md, slash-commands, hooks |
| Triggering | auto-trigger |
| GitHub metrics | 248,154 stars · 22,026 forks · 341 open issues (fetched 2026-07-07) |
| Last commit | 2026-07-06 |
| Created | 2025-10-09 |
Editorial analysis
The most influential community skill framework for Claude Code, and the clearest existing argument that skills should encode process, not just knowledge. Trigger design is unusually good: skills state when they fire and when they must not. Verification-before-completion is the standout artifact. Weaknesses: few skills ship executable tests, and the framework's insistence on skill invocation before any action can add overhead on trivial tasks.
Risks
The using-superpowers meta-skill injects strong mandatory language into every session; teams should review whether skill-priority rules fit their workflow before org-wide adoption.
Evidence notes: Reliability: explicit triggers and red-flag tables in most skills; failure paths documented. Leverage: one install changes behavior across every coding task. Eval readiness: skills describe verification but rarely ship machine-checkable tests. Safety: relies on host permission model; no destructive-action gates of its own.
Install and first run
claude plugin install superpowers@claude-plugins-officialScored skills in this repo
Blocks the agent from claiming work is complete, fixed, or passing until it has run verification commands and confirmed output. Evidence…
Forces root-cause investigation before any fix: reproduce, hypothesize one cause at a time, test the hypothesis, then patch. Bans…
Requires a failing test before implementation code for any feature or bugfix. The loop is red, green, refactor, with the agent forbidden…
Forces requirements exploration before any creative work: intent, constraints, and design alternatives surfaced and agreed before…
Converts a spec into a written implementation plan with ordered steps and checkpoints before any code is touched.
Executes a written plan with review checkpoints: batch-by-batch progress, verification per step, and stop points for feedback.
Governs when and how to fan work out to parallel subagents: independence tests, scoping rules, and result-integration discipline.
Ensures feature work happens in an isolated workspace via git worktrees before executing plans or risky changes.
Requires completed work to pass through a review pass against requirements before merging or declaring done.
Governs how feedback gets handled: technical verification of each finding before implementation, no performative agreement, pushback with…
Executes plan tasks through fresh subagents per task, keeping the orchestrating session's context clean and each task's context focused.
The meta-skill: how to author, edit, and verify skills, including trigger design, structure conventions, and pre-deployment testing.
Structures the end of feature work: verify tests pass, then present explicit merge/PR/cleanup options instead of assuming an integration…
Alternatives in this category
An opinionated 23-skill suite that staffs a solo developer's agent with the roles a startup would hire: CEO review, design review,…
A CLI for configuring Claude Code: install commands, agents, hooks, and settings from a large template catalog, plus usage monitoring.
A configuration framework layering specialized commands and cognitive personas onto Claude Code: an opinionated operating mode rather than a…
Embed this badge
Maintainers of scored entries can embed their live badge; it updates when the score does and links back to this page.
[](https://agentiquette.com/index/repos/superpowers)Frequently asked questions
Is obra/superpowers good?
obra/superpowers scores 83/100 (A-tier) on Agentiquette's methodology v1.0, scored 2026-07-07. Strengths and weaknesses are in the editorial analysis; the score is explained dimension by dimension in the scorecard.
What is obra/superpowers for?
A process-first skill framework for Claude Code: brainstorming, test-driven development, systematic debugging, and verification skills that gate how the agent works, not just what it knows.
Is obra/superpowers maintained?
Last commit 2026-07-06, 341 open issues at fetch time (2026-07-07). Maintenance is scored at 92 of 100.
Maintainer of obra/superpowers? Claim this listing and suggest corrections.