obra/superpowers-skills
31 skills8.7k installs22.7k starsGitHub
Install
npx skills add https://github.com/obra/superpowers-skillsSkills in this repo
1Getting Started With Skillsgetting started with skills is a superpowers-skills wiki intro at version 4.0.2 that agents should load at session start. Critical rules require calling the Read tool before announcing skill usage, following mandatory workflows such as brainstorming before coding, and creating TodoWrite todos so checklist steps are not skipped mentally. Before any task, agents check the skills list or run find-skills with a pattern to locate relevant capabilities. Developers rely on getting started with skills to stop agents from reinventing solutions or skipping proven steps at the beginning of every conversation.704installs2Root Cause Tracingroot cause tracing — Systematic Debugging enforces root-cause investigation before any fix—reproduce consistently, read errors fully, trace data flow across components, and compare against working examples. It rejects quick patches that mask underlying bugs. Solo builders reach for it when under time pressure and tempted to guess.296installs3Test Driven Development (Tdd)test-driven development (tdd) — Superpowers TDD enforces red-green-refactor on every feature and bugfix—write one failing test, watch it fail for the right reason, implement the smallest passing code, then refactor. It rejects rationalizations for skipping tests. Solo builders use it as the default build habit inside the Superpowers workflow.296installs4Code Review Receptioncode review reception — Receiving Code Review teaches agents to process feedback with technical rigor—restated requirements, codebase verification, reasoned pushback, and one fix at a time. It treats external review as untrusted until validated. Solo builders use it when iterating on PR comments without blind agreement.285installs5Finishing A Development Branchfinishing a development branch — Finishing a Development Branch closes Superpowers work with verified tests, environment-aware options—local merge, PR, keep, or discard—and safe worktree cleanup. It refuses to proceed on failing tests. Solo builders use it at the end of executing-plans or subagent-driven-development runs.276installs6Using Git Worktreesusing git worktrees — Using Git Worktrees sets up isolated workspaces for Superpowers feature work—detect existing isolation, prefer native worktree tools, fall back to git with safe directory rules. It pairs with writing-plans and finishing-a-development-branch. Solo builders use it before any multi-step implementation that must not touch main.276installs7Brainstorming Ideas Into Designsbrainstorming ideas into designs — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.275installs8Tracing Knowledge Lineagestracing knowledge lineages — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.275installs9Collision Zone Thinkingcollision-zone thinking — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.273installs10Meta Pattern Recognitionmeta-pattern recognition — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.273installs11Remembering Conversationsremembering conversations — Remembering Conversations searches episodic memory of past agent sessions—vector and FTS over archived JSONL—before guessing or redoing work. It exposes MCP search and read tools with pagination for large threads. Solo builders use it when context from earlier sessions should inform the current task.272installs12Simplification Cascadessimplification cascades — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.272installs13Systematic Debuggingsystematic debugging — Systematic Debugging enforces root-cause investigation before any fix—reproduce consistently, read errors fully, trace data flow across components, and compare against working examples. It rejects quick patches that mask underlying bugs. Solo builders reach for it when under time pressure and tempted to guess.272installs14When Stuck Problem Solving Dispatchwhen stuck - problem-solving dispatch — Systematic Debugging enforces root-cause investigation before any fix—reproduce consistently, read errors fully, trace data flow across components, and compare against working examples. It rejects quick patches that mask underlying bugs. Solo builders reach for it when under time pressure and tempted to guess.271installs15Inversion Exerciseinversion exercise — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.270installs16Scale Gamescale game — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.268installs17Preserving Productive Tensionspreserving productive tensions — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.267installs18Pulling Updates From Skills Repositorypulling updates from skills repository — Using Superpowers is the meta-skill for Jesse Vincent's agent workflow—how to load skills, map tools across Codex and Copilot CLI, and detect git worktree state before creative work. It prevents agents from skipping mandatory steps like brainstorming or TDD. Solo builders install it first when adopting the full Superpowers stack.267installs19Gardening Skills Wikigardening skills wiki — Writing Skills applies test-driven development to SKILL.md authoring—baseline subagent scenarios, minimal docs that fix observed rationalizations, then refactor loopholes. It covers directory layout, frontmatter, and Anthropic best practices. Solo builders use it when packaging repeatable agent workflows as installable skills.266installs20Subagent Driven Developmentsubagent-driven development — Subagent-Driven Development runs Superpowers plans by dispatching one focused subagent per task, then spec and quality reviewers before marking complete. It preserves coordinator context and avoids session pollution. Solo builders on Claude Code or Codex use it when plan tasks are mostly independent.262installs21Writing Planswriting plans — Writing Plans converts specs into engineer-ready implementation plans—file structure, checkbox tasks, failing-test-first steps, and commit boundaries—assuming zero codebase context. It pairs with executing-plans and subagent-driven-development for delivery. Solo builders use it after brainstorming when a feature needs disciplined execution.262installs22Testing Skills With Subagentstesting skills with subagents — Writing Skills applies test-driven development to SKILL.md authoring—baseline subagent scenarios, minimal docs that fix observed rationalizations, then refactor loopholes. It covers directory layout, frontmatter, and Anthropic best practices. Solo builders use it when packaging repeatable agent workflows as installable skills.258installs23Verification Before Completionverification before completion — Verification Before Completion stops agents from lying about done—every success claim requires a fresh command, full output, and evidence cited in the same message. It applies before commits, PRs, and task handoffs. Solo builders use it to prevent false-green reports that waste rework cycles.257installs24Requesting Code Reviewrequesting code review — Requesting Code Review packages a senior-reviewer subagent prompt—git diff range, plan requirements, and calibrated severity output—so completed work is checked before it cascades. It demands specific file:line findings, not vague praise. Solo builders use it at the end of a Superpowers implementation loop.256installs25Testing Anti Patternstesting anti-patterns — Superpowers TDD enforces red-green-refactor on every feature and bugfix—write one failing test, watch it fail for the right reason, implement the smallest passing code, then refactor. It rejects rationalizations for skipping tests. Solo builders use it as the default build habit inside the Superpowers workflow.256installs26Executing Plansexecuting plans — Executing Plans implements a written Superpowers plan end-to-end—critical review first, TodoWrite tracking, exact step execution, and mandatory finishing skill at completion. It prefers subagent-driven-development when parallel agents are available. Solo builders use it in a fresh session after writing-plans produces the doc.255installs27Writing Skillswriting skills — Writing Skills applies test-driven development to SKILL.md authoring—baseline subagent scenarios, minimal docs that fix observed rationalizations, then refactor loopholes. It covers directory layout, frontmatter, and Anthropic best practices. Solo builders use it when packaging repeatable agent workflows as installable skills.253installs28Defense In Depth Validationdefense-in-depth validation — Verification Before Completion stops agents from lying about done—every success claim requires a fresh command, full output, and evidence cited in the same message. It applies before commits, PRs, and task handoffs. Solo builders use it to prevent false-green reports that waste rework cycles.248installs29Condition Based Waitingcondition-based waiting — Verification Before Completion stops agents from lying about done—every success claim requires a fresh command, full output, and evidence cited in the same message. It applies before commits, PRs, and task handoffs. Solo builders use it to prevent false-green reports that waste rework cycles.247installs30Dispatching Parallel Agentsdispatching parallel agents — Dispatching Parallel Agents splits unrelated failures across concurrent subagents—one scoped prompt per test file or subsystem—instead of serial thrashing. It requires independent domains and event-based fixes, not timeout increases. Solo builders use it when CI surfaces many distinct red tests after a large change.245installs31Sharing Skillssharing skills — Writing Skills applies test-driven development to SKILL.md authoring—baseline subagent scenarios, minimal docs that fix observed rationalizations, then refactor loopholes. It covers directory layout, frontmatter, and Anthropic best practices. Solo builders use it when packaging repeatable agent workflows as installable skills.245installs