{"skill":{"slug":"superpowers","displayName":"Superpowers Dev Workflow","summary":"Spec-first, TDD, subagent-driven software development workflow. Use when: (1) building any new feature or app — triggers brainstorm → plan → subagent executi...","description":"---\nname: superpowers\ndescription: >\n  Spec-first, TDD, subagent-driven software development workflow. Use when:\n  (1) building any new feature or app — triggers brainstorm → plan → subagent execution loop,\n  (2) debugging a bug or test failure — triggers systematic root-cause process,\n  (3) user says \"let's build\", \"help me plan\", \"I want to add X\", or \"this is broken\",\n  (4) completing a feature branch — triggers test verification + merge/PR options.\n  NOT for: one-liner fixes (just edit), reading code, or non-code tasks.\n  Requires exec tool and sessions_spawn.\n---\n\n# Superpowers — OpenClaw Edition\n\nAdapted from [obra/superpowers](https://github.com/obra/superpowers). Mandatory workflow — not suggestions.\n\n## The Pipeline\n\n```\nIdea → Brainstorm → Plan → Subagent-Driven Build (TDD) → Code Review → Finish Branch\n```\n\nEvery coding task follows this pipeline. \"Too simple to need a design\" is always wrong.\n\n---\n\n## Phase 1: Brainstorming\n\n**Trigger:** User wants to build something. Activate before touching any code.\n\n**See:** [references/brainstorming.md](references/brainstorming.md)\n\n**Summary:**\n1. Explore project context (files, docs, recent commits)\n2. Ask clarifying questions — **one at a time**, prefer multiple choice\n3. Propose 2–3 approaches with trade-offs + recommendation\n4. Present design in sections, get approval after each\n5. Write design doc → `docs/plans/YYYY-MM-DD-<topic>-design.md` → commit\n6. Hand off to **Phase 2: Writing Plans**\n\n**HARD GATE:** Do NOT write any code until user approves design.\n\n---\n\n## Phase 2: Writing Plans\n\n**Trigger:** Design approved. Activated by brainstorming phase.\n\n**See:** [references/writing-plans.md](references/writing-plans.md)\n\n**Summary:**\n- Write a detailed task-by-task implementation plan\n- Each task = 2–5 minutes: write test → watch fail → implement → watch pass → commit\n- Save to `docs/plans/YYYY-MM-DD-<feature>.md`\n- Announce: `\"I'm using the writing-plans skill to create the implementation plan.\"`\n- After saving, offer two execution modes:\n  - **Subagent-driven (current session):** `sessions_spawn` per task + two-stage review\n  - **Manual execution:** User runs tasks themselves\n\n---\n\n## Phase 3: Subagent-Driven Development\n\n**Trigger:** Plan exists, user chooses subagent-driven execution.\n\n**See:** [references/subagent-development.md](references/subagent-development.md)\n\n**Per-task loop (OpenClaw):**\n1. `sessions_spawn` an implementer subagent with task + full plan context\n2. Wait for completion announcement\n3. `sessions_spawn` a spec-reviewer subagent → must confirm code matches spec\n4. `sessions_spawn` a code-quality reviewer subagent → must approve quality\n5. Fix any issues, re-review if needed\n6. Mark task done, move to next\n7. Final: dispatch overall code reviewer → hand off to Phase 5\n\n**TDD is mandatory in every task.** See [references/tdd.md](references/tdd.md).\n\n---\n\n## Phase 4: Systematic Debugging\n\n**Trigger:** Bug, test failure, unexpected behaviour — any technical issue.\n\n**See:** [references/systematic-debugging.md](references/systematic-debugging.md)\n\n**HARD GATE:** No fixes without root cause investigation first.\n\n**Four phases:**\n1. Root Cause Investigation (read errors, reproduce, check recent changes, trace data flow)\n2. Pattern Analysis (find working examples, compare, identify differences)\n3. Hypothesis + Testing (one hypothesis at a time, test to prove/disprove)\n4. Fix + Verification (fix at root, not symptom; verify fix doesn't break anything)\n\n---\n\n## Phase 5: Finishing a Branch\n\n**Trigger:** All tasks complete, all tests pass.\n\n**See:** [references/finishing-branch.md](references/finishing-branch.md)\n\n**Summary:**\n1. Verify all tests pass\n2. Determine base branch\n3. Present 4 options: merge locally / push + PR / keep / discard\n4. Execute choice\n5. Clean up\n\n---\n\n## OpenClaw Subagent Dispatch Pattern\n\nWhen dispatching implementer or reviewer subagents, use `sessions_spawn`:\n\n```\nGoal: [one sentence]\nContext: [why it matters, which plan file]\nFiles: [exact paths]\nConstraints: [what NOT to do — no scope creep, TDD only]\nVerify: [how to confirm success — tests pass, specific command]\nTask text: [paste full task from plan]\n```\n\nRun `sessions_spawn` with the task as a detailed prompt. The sub-agent announces results automatically.\n\n---\n\n## Key Principles\n\n- **One question at a time** during brainstorm\n- **TDD always** — write failing test first, delete code written before tests\n- **YAGNI** — remove unnecessary features from all designs\n- **DRY** — no duplication\n- **Systematic over ad-hoc** — follow the process especially under time pressure\n- **Evidence over claims** — verify before declaring success\n- **Frequent commits** — after each green test\n","topics":["Debugging","Subagent","Software Development"],"tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":19237,"installsAllTime":657,"installsCurrent":152,"stars":18,"versions":1},"createdAt":1772227064279,"updatedAt":1778993751796},"latestVersion":{"version":"1.0.0","createdAt":1772227064279,"changelog":"Initial port of obra/superpowers for OpenClaw — spec-first, TDD, subagent-driven development workflow","license":null},"metadata":null,"owner":{"handle":"wlshlad85","userId":"s17f2yfxj46m8sn9jyk0tvm0z9884wpk","displayName":"wlshlad85","image":"https://avatars.githubusercontent.com/u/146856731?v=4"},"moderation":null}