{"id":4140,"date":"2026-08-18T15:03:01","date_gmt":"2026-08-18T07:03:01","guid":{"rendered":"https:\/\/ai.wuaishare.cn\/hub\/?p=4140"},"modified":"2026-08-18T15:03:01","modified_gmt":"2026-08-18T07:03:01","slug":"superpowers","status":"publish","type":"post","link":"https:\/\/ai.wuaishare.cn\/hub\/en\/4140","title":{"rendered":"Superpowers"},"content":{"rendered":"<p># Superpowers<\/p>\n<p>Superpowers is a complete software development methodology for your coding agents, built on top of a set of composable skills and some initial instructions that make sure your agent uses them.<\/p>\n<p>## Table of Contents<\/p>\n<p>&#8211; [How it works](#how-it-works)<br \/>\n&#8211; [Commercial Services](#commercial-services)<br \/>\n&#8211; [Getting Started](#installation)<br \/>\n  &#8211; [Claude Code](#claude-code)<br \/>\n  &#8211; [Antigravity](#antigravity)<br \/>\n  &#8211; [Codex App](#codex-app)<br \/>\n  &#8211; [Codex CLI](#codex-cli)<br \/>\n  &#8211; [Cursor](#cursor)<br \/>\n  &#8211; [Devin CLI](#devin-cli)<br \/>\n  &#8211; [Factory Droid](#factory-droid)<br \/>\n  &#8211; [Gemini CLI](#gemini-cli)<br \/>\n  &#8211; [GitHub Copilot CLI](#github-copilot-cli)<br \/>\n  &#8211; [Grok Build CLI](#grok-build-cli)<br \/>\n  &#8211; [Kimi Code](#kimi-code)<br \/>\n  &#8211; [OpenCode](#opencode)<br \/>\n  &#8211; [Pi](#pi)<br \/>\n  &#8211; [Hermes Agent](#hermes-agent)<br \/>\n&#8211; [The Basic Workflow](#the-basic-workflow)<br \/>\n&#8211; [Community](#community)<br \/>\n&#8211; [What&#8217;s Inside](#whats-inside)<br \/>\n&#8211; [Philosophy](#philosophy)<br \/>\n&#8211; [Contributing](#contributing)<br \/>\n&#8211; [Updating](#updating)<br \/>\n&#8211; [License](#license)<br \/>\n&#8211; [Visual companion telemetry](#visual-companion-telemetry)<\/p>\n<p>## How it works<\/p>\n<p>It starts from the moment you fire up your coding agent. As soon as it sees that you&#8217;re building something, it *doesn&#8217;t* just jump into trying to write code. Instead, it steps back and asks you what you&#8217;re really trying to do. <\/p>\n<p>Once it&#8217;s teased a spec out of the conversation, it shows it to you in chunks short enough to actually read and digest. <\/p>\n<p>After you&#8217;ve signed off on the design, your agent puts together an implementation plan that&#8217;s clear enough for an enthusiastic junior engineer with poor taste, no judgement, no project context, and an aversion to testing to follow. It emphasizes true red\/green TDD, YAGNI (You Aren&#8217;t Gonna Need It), and DRY. <\/p>\n<p>Next up, once you say &#8220;go&#8221;, it launches a *subagent-driven-development* process, having agents work through each engineering task, inspecting and reviewing their work, and continuing forward. It&#8217;s not uncommon for your agent to work autonomously for a couple hours at a time without deviating from the plan you put together.<\/p>\n<p>There&#8217;s a bunch more to it, but that&#8217;s the core of the system. And because the skills trigger automatically, you don&#8217;t need to do anything special. Your coding agent just has Superpowers.<\/p>\n<p>## Commercial Services<\/p>\n<p>If you&#8217;re using Superpowers in enterprise and could benefit from commercial support, additional tooling, or managed spending, please don&#8217;t hesitate to drop us a line at sales@primeradiant.com.<\/p>\n<p>## Installation<\/p>\n<p>Installation differs by harness. If you use more than one, install Superpowers separately for each one.<\/p>\n<p>### Claude Code<\/p>\n<p>Superpowers is available via the [official Claude plugin marketplace](https:\/\/claude.com\/plugins\/superpowers)<\/p>\n<p>#### Official Marketplace<\/p>\n<p>&#8211; Install the plugin from Anthropic&#8217;s official marketplace:<\/p>\n<p>  &#8220;`bash<br \/>\n  \/plugin install superpowers@claude-plugins-official<br \/>\n  &#8220;`<\/p>\n<p>#### Superpowers Marketplace<\/p>\n<p>The Superpowers marketplace provides Superpowers and some other related plugins for Claude Code.<\/p>\n<p>&#8211; Register the marketplace:<\/p>\n<p>  &#8220;`bash<br \/>\n  \/plugin marketplace add obra\/superpowers-marketplace<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Install the plugin from this marketplace:<\/p>\n<p>  &#8220;`bash<br \/>\n  \/plugin install superpowers@superpowers-marketplace<br \/>\n  &#8220;`<\/p>\n<p>### Antigravity<\/p>\n<p>Install Superpowers as a plugin from this repository:<\/p>\n<p>&#8220;`bash<br \/>\nagy plugin install https:\/\/github.com\/obra\/superpowers<br \/>\n&#8220;`<\/p>\n<p>Antigravity runs the plugin&#8217;s session-start hook, so Superpowers is active from<br \/>\nthe first message. Reinstall with the same command to update.<\/p>\n<p>### Codex App<\/p>\n<p>Superpowers is available via the [official Codex plugin marketplace](https:\/\/github.com\/openai\/plugins).<\/p>\n<p>&#8211; In the Codex app, click on Plugins in the sidebar.<br \/>\n&#8211; You should see `Superpowers` in the Coding section.<br \/>\n&#8211; Click the `+` next to Superpowers and follow the prompts.<\/p>\n<p>### Codex CLI<\/p>\n<p>Superpowers is available via the [official Codex plugin marketplace](https:\/\/github.com\/openai\/plugins).<\/p>\n<p>&#8211; Open the plugin search interface:<\/p>\n<p>  &#8220;`bash<br \/>\n  \/plugins<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Search for Superpowers:<\/p>\n<p>  &#8220;`bash<br \/>\n  superpowers<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Select `Install Plugin`.<\/p>\n<p>### Cursor<\/p>\n<p>&#8211; In Cursor Agent chat, install from marketplace:<\/p>\n<p>  &#8220;`text<br \/>\n  \/add-plugin superpowers<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Or search for &#8220;superpowers&#8221; in the plugin marketplace.<\/p>\n<p>### Devin CLI<\/p>\n<p>&#8211; Install the plugin from this repository:<\/p>\n<p>  &#8220;`bash<br \/>\n  devin plugins install obra\/superpowers<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Update to the latest version with:<\/p>\n<p>  &#8220;`bash<br \/>\n  devin plugins update superpowers<br \/>\n  &#8220;`<\/p>\n<p>### Factory Droid<\/p>\n<p>&#8211; Register the marketplace:<\/p>\n<p>  &#8220;`bash<br \/>\n  droid plugin marketplace add https:\/\/github.com\/obra\/superpowers<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Install the plugin:<\/p>\n<p>  &#8220;`bash<br \/>\n  droid plugin install superpowers@superpowers<br \/>\n  &#8220;`<\/p>\n<p>### Gemini CLI<\/p>\n<p>&#8211; Install the extension:<\/p>\n<p>  &#8220;`bash<br \/>\n  gemini extensions install https:\/\/github.com\/obra\/superpowers<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Update later:<\/p>\n<p>  &#8220;`bash<br \/>\n  gemini extensions update superpowers<br \/>\n  &#8220;`<\/p>\n<p>### GitHub Copilot CLI<\/p>\n<p>&#8211; Register the marketplace:<\/p>\n<p>  &#8220;`bash<br \/>\n  copilot plugin marketplace add obra\/superpowers-marketplace<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Install the plugin:<\/p>\n<p>  &#8220;`bash<br \/>\n  copilot plugin install superpowers@superpowers-marketplace<br \/>\n  &#8220;`<\/p>\n<p>### Grok Build CLI<\/p>\n<p>Superpowers is available via the [official Grok plugin marketplace](https:\/\/github.com\/xai-org\/plugin-marketplace).<\/p>\n<p>&#8211; Install the plugin from xAI&#8217;s official marketplace:<\/p>\n<p>  &#8220;`bash<br \/>\n  grok plugin install superpowers@xai-official &#8211;trust<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Or open the marketplace in the TUI, search for Superpowers, and install it:<\/p>\n<p>  &#8220;`text<br \/>\n  \/marketplace<br \/>\n  &#8220;`<\/p>\n<p>### Kimi Code<\/p>\n<p>Superpowers is available in Kimi Code&#8217;s plugin marketplace.<\/p>\n<p>&#8211; Open Kimi Code&#8217;s plugin manager:<\/p>\n<p>  &#8220;`text<br \/>\n  \/plugins<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Go to `Marketplace` > `Superpowers` and install it.<\/p>\n<p>&#8211; Or install directly from this repository:<\/p>\n<p>  &#8220;`text<br \/>\n  \/plugins install https:\/\/github.com\/obra\/superpowers<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Detailed docs: [docs\/README.kimi.md](docs\/README.kimi.md)<\/p>\n<p>### OpenCode<\/p>\n<p>OpenCode uses its own plugin install; install Superpowers separately even if you<br \/>\nalready use it in another harness.<\/p>\n<p>&#8211; Tell OpenCode:<\/p>\n<p>  &#8220;`<br \/>\n  Fetch and follow instructions from https:\/\/raw.githubusercontent.com\/obra\/superpowers\/refs\/heads\/main\/.opencode\/INSTALL.md<br \/>\n  &#8220;`<\/p>\n<p>&#8211; Detailed docs: [docs\/README.opencode.md](docs\/README.opencode.md)<\/p>\n<p>### Pi<\/p>\n<p>Install Superpowers as a Pi package from this repository:<\/p>\n<p>&#8220;`bash<br \/>\npi install git:github.com\/obra\/superpowers<br \/>\n&#8220;`<\/p>\n<p>For local development, run Pi with this checkout loaded as a temporary package:<\/p>\n<p>&#8220;`bash<br \/>\npi -e \/path\/to\/superpowers<br \/>\n&#8220;`<\/p>\n<p>The Pi package loads the Superpowers skills and a small extension that injects the `using-superpowers` bootstrap at session startup and again after compaction. Pi has native skills, so no compatibility `Skill` tool is required. Subagent and task-list tools remain optional Pi companion packages.<\/p>\n<p>### Hermes Agent<\/p>\n<p>Install Superpowers as a Hermes plugin from this repository:<\/p>\n<p>&#8220;`bash<br \/>\nhermes plugins install obra\/superpowers &#8211;enable<br \/>\n&#8220;`<\/p>\n<p>Restart any active Hermes sessions after installing. Note: Hermes has no<br \/>\npost-compaction hook, so a very long session that compacts over its first<br \/>\nturn loses the bootstrap \u2014 start a fresh session if skills stop triggering.<\/p>\n<p>## The Basic Workflow<\/p>\n<p>1. **brainstorming** &#8211; Activates before writing code. Refines rough ideas through questions, explores alternatives, presents design in sections for validation. Saves design document.<\/p>\n<p>2. **using-git-worktrees** &#8211; Activates after design approval. Creates isolated workspace on new branch, runs project setup, verifies clean test baseline.<\/p>\n<p>3. **writing-plans** &#8211; Activates with approved design. Breaks work into bite-sized tasks (2-5 minutes each). Every task has exact file paths, complete code, verification steps.<\/p>\n<p>4. **subagent-driven-development** or **executing-plans** &#8211; Activates with plan. Dispatches fresh subagent per task with two-stage review (spec compliance, then code quality), or executes in batches with human checkpoints.<\/p>\n<p>5. **test-driven-development** &#8211; Activates during implementation. Enforces RED-GREEN-REFACTOR: write failing test, watch it fail, write minimal code, watch it pass, commit. Deletes code written before tests.<\/p>\n<p>6. **requesting-code-review** &#8211; Activates between tasks. Reviews against plan, reports issues by severity. Critical issues block progress.<\/p>\n<p>7. **finishing-a-development-branch** &#8211; Activates when tasks complete. Verifies tests, presents options (merge\/PR\/keep\/discard), cleans up worktree.<\/p>\n<p>**The agent checks for relevant skills before any task.** Mandatory workflows, not suggestions.<\/p>\n<p>## Community<\/p>\n<p>Superpowers is built by [Jesse Vincent](https:\/\/blog.fsck.com) and the rest of the folks at [Prime Radiant](https:\/\/primeradiant.com).<\/p>\n<p>&#8211; **Discord**: [Join us](https:\/\/discord.gg\/35wsABTejz) for community support, questions, and sharing what you&#8217;re building with Superpowers<br \/>\n&#8211; **Issues**: https:\/\/github.com\/obra\/superpowers\/issues<br \/>\n&#8211; **Release announcements**: [Sign up](https:\/\/primeradiant.com\/superpowers\/) to get notified about new versions<\/p>\n<p>## What&#8217;s Inside<\/p>\n<p>### Skills Library<\/p>\n<p>**Testing**<br \/>\n&#8211; **test-driven-development** &#8211; RED-GREEN-REFACTOR cycle (includes testing anti-patterns reference)<\/p>\n<p>**Debugging**<br \/>\n&#8211; **systematic-debugging** &#8211; 4-phase root cause process (includes root-cause-tracing, defense-in-depth, condition-based-waiting techniques)<br \/>\n&#8211; **verification-before-completion** &#8211; Ensure it&#8217;s actually fixed<\/p>\n<p>**Collaboration**<br \/>\n&#8211; **brainstorming** &#8211; Socratic design refinement<br \/>\n&#8211; **writing-plans** &#8211; Detailed implementation plans<br \/>\n&#8211; **executing-plans** &#8211; Batch execution with checkpoints<br \/>\n&#8211; **dispatching-parallel-agents** &#8211; Concurrent subagent workflows<br \/>\n&#8211; **requesting-code-review** &#8211; Pre-review checklist<br \/>\n&#8211; **receiving-code-review** &#8211; Responding to feedback<br \/>\n&#8211; **using-git-worktrees** &#8211; Parallel development branches<br \/>\n&#8211; **finishing-a-development-branch** &#8211; Merge\/PR decision workflow<br \/>\n&#8211; **subagent-driven-development** &#8211; Fast iteration with two-stage review (spec compliance, then code quality)<\/p>\n<p>**Meta**<br \/>\n&#8211; **writing-skills** &#8211; Create new skills following best practices (includes testing methodology)<br \/>\n&#8211; **using-superpowers** &#8211; Introduction to the skills system<\/p>\n<p>## Philosophy<\/p>\n<p>&#8211; **Test-Driven Development** &#8211; Write tests first, always<br \/>\n&#8211; **Systematic over ad-hoc** &#8211; Process over guessing<br \/>\n&#8211; **Complexity reduction** &#8211; Simplicity as primary goal<br \/>\n&#8211; **Evidence over claims** &#8211; Verify before declaring success<\/p>\n<p>Read [the original release announcement](https:\/\/blog.fsck.com\/2025\/10\/09\/superpowers\/).<\/p>\n<p>## Contributing<\/p>\n<p>The general contribution process for Superpowers is below. Keep in mind that we don&#8217;t generally accept contributions of new skills and that any updates to skills must work across all of the coding agents we support.<\/p>\n<p>1. Fork the repository<br \/>\n2. Switch to the &#8216;dev&#8217; branch<br \/>\n3. Create a branch for your work<br \/>\n4. Follow the `writing-skills` skill for creating and testing new and modified skills<br \/>\n5. Submit a PR, being sure to fill in the pull request template.<\/p>\n<p>Skill-behavior tests use the drill eval harness from [superpowers-evals](https:\/\/github.com\/prime-radiant-inc\/superpowers-evals\/), cloned into `evals\/` \u2014 see `evals\/README.md` for setup. Plugin-infrastructure tests live at `tests\/` and run via the relevant `run-*.sh` or `npm test`.<\/p>\n<p>See `skills\/writing-skills\/SKILL.md` for the complete guide.<\/p>\n<p>## Updating<\/p>\n<p>Superpowers updates are somewhat coding-agent dependent, but are often automatic.<\/p>\n<p>## License<\/p>\n<p>MIT License &#8211; see LICENSE file for details<\/p>\n<p>## Visual companion telemetry<\/p>\n<p>Because skills and plugins don&#8217;t provide any feedback to creators, we have no idea how many of you are using Superpowers. By default, the Prime Radiant logo on brainstorming&#8217;s optional visual companion feature is loaded from our website. It includes the version of Superpowers in use. It does not include any details about your project, prompt, or coding agent. We don&#8217;t see your clicks or anything about what you&#8217;re building. This helps us have a rough idea of how many folks are using Superpowers and which version of Superpowers they&#8217;re using. It&#8217;s 100% optional. To disable this, set the environment variable `SUPERPOWERS_DISABLE_TELEMETRY` to any true value. Superpowers also honors Claude Code&#8217;s `DISABLE_TELEMETRY` and `CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC` opt-outs.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>An agentic skills framework &#038; software development methodology that works.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_share_source_url":"https:\/\/github.com\/obra\/superpowers","ai_share_source_title":"obra\/superpowers","ai_share_status":"\u5df2\u53d1\u5e03","ai_share_original_content_language":"en","ai_share_i18n_original_source_language":"en","ai_share_author_label":"obra","ai_share_version_label":"v6.3.0","ai_share_primary_category":"\u7f16\u7a0b\u5f00\u53d1","ai_share_tool_type":"skill","ai_share_resource_type":"workflow_skill_pack","ai_share_platform_support":"CLI,\u6587\u6863","ai_share_collection":"AI \u7f16\u7a0b\u5de5\u4f5c\u53f0","ai_share_last_verified_at":"2026-05-08","ai_share_provider_observed_at":"2026-08-18T06:51:09.387Z","ai_share_last_commit_at":"2026-08-13T00:36:31Z","ai_share_license":"MIT","ai_share_language_support":"en","ai_share_target_user":"AI coding \u5f00\u53d1\u8005,\u5de5\u7a0b\u56e2\u961f","ai_share_github_topics":"ai, brainstorming, coding, obra, sdlc, skills, subagent-driven-development, superpowers","ai_share_repo_default_branch":"main","ai_share_pricing_model":"open_source","ai_share_deployment_mode":"self_hosted","ai_share_delivery_mode":"","ai_share_region_availability":"global","ai_share_source_platform":"GitHub","ai_share_release_url":"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.3.0","ai_share_github_releases_json":"[{\"tagName\":\"v6.3.0\",\"name\":\"v6.3.0\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.3.0\",\"body\":\"### Harness Support\\n\\n- **Devin CLI**: `devin plugins install obra\/superpowers` now works, and skills auto-trigger at session start. (#1995)\\n- **Hermes Agent**: install from a git clone; skills register with Hermes' native loader and the bootstrap loads on the first turn. (#1922, #2025)\\n- **Grok Build CLI** added to the install docs. (#1919)\\n\\n### Brainstorming\\n\\n- **Ceremony now scales to the task.** Requests are classified as spike, bounded, or architectural; small tasks skip the two-document ritual. Every path still stops for your approval before implementation. (#2063)\\n\\n### Subagent-Driven Development\\n\\n- **Controllers no longer stall on plan conflicts.** Non-catastrophic conflicts and ambiguities get a recorded ruling and work continues; only destructive or irreversible actions still stop for a human. One donated session had sat blocked for almost nine hours on a question the controller could have decided. (#2077)\\n- **The pre-dispatch conflict scan records its checks in the ledger** instead of just asserting the plan is clean. (#2080)\\n- **Small same-shape tasks batch into one dispatch**, cutting subagent cost sharply on micro-task plans; batch reviews verify every file in the brief made it into the diff. (#2078)\\n- **Implementers and reviewers may not spawn their own subagents**, which was producing duplicate reviews. (#2059)\\n- **Plans carry a `Spec:` pointer** and SDD reads the spec at setup, so plan conflicts get resolved against the design instead of guessed at. (#2086)\\n- Reviewers re-read evidence they find illegible instead of re-running the test suite (#2089), and circuit-breaker rulings now show up in the Finish report.\\n\\n### Codex\\n\\n- Subagent waits are event-driven instead of poll-heavy, spawns pin model and reasoning effort explicitly, and the multi-agent reference is corrected against Codex source. (#2060, #2061, #2062)\\n\\n### Finishing a Development Branch\\n\\n- **Worktree removal no longer destroys untracked files.** When `git worktree remove` refuses because the tree holds uncommitted work, the skill stops, names the files, and asks \u2014 instead of reaching for `--force`. (#2016, #1223, #2024)\\n\\n### Fixes\\n\\n- `render-graphs.js` in writing-skills works on Windows.\\n- Corrected Copilot CLI backgrounding guidance for Windows. (#1929, #2006)\\n- `bump-version.sh` covers the Hermes manifest.\\n\\n### Documentation\\n\\n- README: added a table of contents and reorganized Getting Started.\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-08-12T16:58:30Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v6.2.0\",\"name\":\"v6.2.0\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.2.0\",\"body\":\"## v6.2.0 (2026-07-23)\\n\\n### Subagent-Driven Development\\n\\nTwo structural changes to how SDD tracks progress and closes out review findings, both developed against live eval campaigns.\\n\\n- **The workspace is now plan-scoped.** `.superpowers\/sdd\/` had no plan identity and no end-of-life: a follow-up plan in the same working tree could read the previous plan's ledger as its own progress (observed in the wild, with multiple contamination rounds and ad-hoc workarounds). `sdd-workspace` now requires the plan file and resolves a per-plan directory, `.superpowers\/sdd\/<plan-basename>\/`; `task-brief` and `review-package` write into their plan's directory (`review-package` gains the plan file as its first argument); the ledger names its plan on its first line; and the workspace is deleted once the final review is clean \u2014 git history is the durable record. Baseline evals showed controllers already refused foreign ledgers, but at a cost of 6\u201313 tool calls of cross-plan git forensics per resume; plan-scoping makes the answer structural instead. (25\/25 baseline and GREEN eval runs documented in `docs\/specs\/` and `docs\/plans\/`.)\\n- **The review-fix loop resumes the implementer.** The lifecycle restructure gives fix rounds resume-the-implementer semantics instead of fresh dispatches, adds a scoped re-review prompt (`re-review-prompt.md`) so the re-reviewer checks the fixes rather than re-reading the whole task, and installs a five-round circuit breaker with controller adjudication when it trips. SKILL.md reorganizes by lifecycle, and its Red Flags convert to the house rationalization-table form.\\n\\n### Skills\\n\\nA branch-wide compression campaign: recap sections, social proof, and benefits-selling prose aimed at a reader who has already invoked the skill are gone, with every load-bearing argument folded into a rationalization-table row or moved to its point of use. Each cut was micro-tested with subagent probes, and the one cut that measurably degraded behavior was reworked rather than shipped.\\n\\n- **`testing-anti-patterns.md` is now `writing-good-tests.md`.** The TDD reference doc is rebuilt as a positive catalog \u2014 six rules that lead with the GOOD example \u2014 and absorbs a falsifiability discipline: name the production change that would fail the test, derive expectations independently of the code under test, and a closing mutation check. It closes two holes by name: the string-presence trap (grep-style tests on scripts, skills, and prompts counterfeit falsifiability \u2014 the observable is behavior, never text) and the change-detector trap (a constant assertion can fail and still protect nothing), each with a hard stop in the gate function. Trivial code and human prose earn no test; the trigger broadens from \\\"adding mocks\\\" to any test writing.\\n- **TDD's \\\"Why Order Matters\\\" rebuttals survive as rationalization rows.** Deleting the section outright measurably degraded test-first behavior under \\\"just write it, tests after\\\" pressure (control 8\/10 \u2192 treatment 5\/10, corroborated on Claude and Codex), so each prose rebuttal now lives in its Common Rationalizations row \u2014 the section is gone but the arguments fire where an agent hits them mid-rationalization.\\n- **`finishing-a-development-branch` no longer offers to discard your work.** The completion menu dates from when throwing away branches was routine; \\\"Discard this work\\\" next to \\\"Merge\\\" advertised destroying finished, passing work. Discard survives as an explicit-request-only path with the same typed-confirmation ritual. The same pass made PR creation forge-agnostic (your forge's CLI or the URL printed on push, not a blessed list of tools) and fixed a real bug: the worktree path was recomputed after cleanup had already changed directory, so provenance checks never matched and cleanup silently no-oped.\\n- **Recap and persuasion prose removed across the library.** `brainstorming`, `systematic-debugging`, `dispatching-parallel-agents`, `verification-before-completion`, `executing-plans`, `subagent-driven-development`, `requesting-code-review`, `receiving-code-review`, `using-git-worktrees`, `writing-plans`, and `writing-skills` all drop their Bottom Line \/ Key Principles \/ Real-World Impact \/ Advantages sections; `using-git-worktrees` and `finishing-a-development-branch` convert their guard sections to the house Excuse\/Reality rationalization table.\\n\\n### Windows\\n\\n- **The SessionStart hook now dispatches via Git Bash.** The hook's command string starts with a quoted path, which broke both shells Claude Code might hand it to: PowerShell parsed the quoted string as an expression and died with a parser error (#1751), and cmd.exe's quote-stripping rule truncated the command when the profile path contained a metacharacter like `(` (#1918) \u2014 either way the bootstrap silently never loaded. The hook now declares `shell: \\\"bash\\\"`, which Claude Code \u2265 2.1.81 resolves to Git for Windows directly, and which surfaces an actionable install prompt when Git Bash is missing. Older Claude Code versions ignore the unknown key and behave as before. Verified end-to-end on Linux, Windows 11 with Git Bash under a hostile path, and Windows 11 without Git Bash.\\n\\n### Harness Support\\n\\n- **Gemini CLI support is restored.** The v6.1.0 removal (on the news that Google had EOLed the Gemini CLI) was premature; the install docs and the `gemini-tools.md` tool-mapping reference are back while permanent removal gets a proper evaluation. (#1959)\\n\\n### Fixes\\n\\n- **`find-polluter.sh` actually finds test files now.** `find .` emits `.\/`-prefixed paths, so the documented `-path \\\"src\/**\/*.test.ts\\\"` pattern matched nothing \u2014 and `wc -l` on empty input then reported \\\"Found 1\\\". Fixed the prefix mismatch (#2008, #2011), plus two follow-ups: a caller-supplied `.\/`-prefixed pattern no longer double-prefixes into a never-matching form, and `**\/` is also matched collapsed so tests directly under the base directory (`src\/top.test.ts` vs `src\/**\/*.test.ts`) aren't silently skipped. The script gains a deterministic test suite.\\n- **The Codex package script works beyond macOS.** Deterministic-metadata tar flags were bsdtar-only spellings, staged file modes depended on two umasks canceling out, and the test's timestamp assertion parsed bsdtar's column layout in a US timezone. GNU tar now gets equivalent flags producing byte-identical headers, modes are pinned canonical, and the test asserts mtime via `tarfile`.\\n- **SDD's skill test no longer flakes.** The file's worst case exceeded the runner's per-file ceiling (raised to 900s), and the assert helpers matched free-form model prose case-sensitively; matching is now case-insensitive and `assert_order` dumps output on failure so the next flake is diagnosable.\\n- **Docs and test cleanup after the v6.1.0 reference pruning.** Dead links to the deleted `claude-code-tools.md`\/`copilot-tools.md` are replaced with the current architecture (#1969), a dangling `#subagent-support` anchor in the Antigravity reference is dropped (#2010), and the Antigravity\/Pi mapping tests assert only the surviving harness-specific mappings \u2014 scoped to the table so they fail again if it's deleted.\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-07-24T00:28:17Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v6.1.1\",\"name\":\"v6.1.1\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.1.1\",\"body\":\"## v6.1.1 (2026-07-02)\\n\\n### Codex\\n\\n- **Codex no longer re-registers the Claude SessionStart hook.** v6.1.0 removed the Codex hook config and its manifest `hooks` pointer, meaning to stop Codex from installing a SessionStart hook \u2014 but with no `hooks` field, Codex fell back to auto-discovering `hooks\/hooks.json`, the Claude Code SessionStart hook that the marketplace ships from the repo root, and re-registered it along with its install-time trust prompt. The Codex manifest now declares an explicit empty hooks object (`hooks: {}`), which Codex reads as \\\"no hooks\\\" instead of reaching the auto-discovery fallback. An absent field, `[]`, and an empty inline list all collapse back to the fallback, so the value has to be exactly `{}`.\\n- **Removed orphaned Codex session-start dead code.** `hooks\/session-start-codex` had no caller once the Codex hook config was deleted, so it and its redundant test cases are gone. The worked shell-hook example in `docs\/porting-to-a-new-harness.md` moves from Codex \u2014 now native skill discovery with no session-start hook \u2014 to Cursor, a live shell-hook harness, and the stale `hooks-codex.json` pointer in `docs\/windows\/polyglot-hooks.md` is corrected. The Codex plugin category is also fixed to \\\"Developer Tools\\\".\\n\\n### Packaging\\n\\n- **New `package-codex-plugin.sh` for building the Codex portal package.** A maintainer script produces a deterministic Codex \\\"portal\\\" archive \u2014 `.zip` by default, `tar.gz` on request \u2014 that normalizes entry timestamps, preserves executable modes, verifies every packaged skill ships its OpenAI metadata, includes the app and composer icons, and refuses to run against a dirty worktree. The packaged manifest keeps the source `hooks: {}` object so a portal-installed plugin avoids the same SessionStart auto-discovery, and the script can rebuild a byte-identical archive from a saved metadata source. Covered by a new test suite.\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-07-02T21:58:30Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v6.1.0\",\"name\":\"v6.1.0\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.1.0\",\"body\":\"### Lower Per-Session Token Cost\\n\\nThe `using-superpowers` bootstrap is injected into every session, so its size is paid for constantly. This release trims it and the per-harness references it points to, without dropping behavior-shaping content.\\n\\n- **Compressed the `using-superpowers` bootstrap.** Replaced the graphviz skill-flow diagram with the prose it encoded, folded the standalone Instruction-Priority section into User Instructions, dropped the per-platform \\\"How to Access Skills\\\" walkthrough, and trimmed the Platform Adaptation pointer to the harnesses that still ship a reference file. The full Red Flags rationalization table and the user-instruction precedence rules are unchanged.\\n- **Pruned the per-harness tool-mapping references.** The verbose action-to-tool tables restated guidance modern agents already follow. Each reference file is trimmed to the harness-specific notes that still carry weight \u2014 subagent dispatch, task tracking, instructions-file paths \u2014 and `claude-code-tools.md` and `copilot-tools.md`, which had nothing harness-specific left, are deleted.\\n\\n### Codex\\n\\n- **Codex can install from the marketplace.** Codex marketplace sources expect a `.agents\/plugins\/marketplace.json` at the marketplace root; the repo only shipped the Claude marketplace file, so Codex could name the marketplace but found no installable plugin entries. A repo-local Codex marketplace manifest now points at the same repository root, so the plugin is installable from Codex.\\n- **Codex no longer ships a SessionStart hook.** Codex reliably triggers skills on its own, and the bootstrap hook made the UX worse rather than better. The Codex hook config (`hooks-codex.json`) and its manifest registration are removed.\\n\\n### Harness Support\\n\\n- **Gemini CLI support removed.** Google EOLed the Gemini CLI on 2026-06-18; the extension can no longer be installed or updated. Gemini is gone from the install docs, the subagent-capable platform lists, and the eval-harness description, and its tool-mapping reference is deleted.\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-30T18:42:18Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v6.0.3\",\"name\":\"v6.0.3\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.0.3\",\"body\":\"### Subagent-Driven Development\\n\\n- **SDD scratch files moved out of `.git\/`.** Claude Code treats `.git\/` as a protected path and denies agent writes there, so an implementer subagent writing its report into `.git\/sdd\/` got blocked mid-run. Task briefs, implementer reports, review diffs, and the progress ledger now live in a self-ignoring `.superpowers\/sdd\/` directory in the working tree \u2014 kept out of `git status` and out of commits, and resolved per worktree by a shared `sdd-workspace` helper. One caveat: because the workspace is git-ignored working-tree scratch, `git clean -fdx` will delete the progress ledger; recover from `git log` if that happens. (#1780)\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-18T22:45:19Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v6.0.2\",\"name\":\"v6.0.2\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.0.2\",\"body\":\"### Install Fixes\\n\\n- **We no longer ship the `evals` submodule.** It broke plugin installs for some users, so the eval harness now lives in its own repo, separate from the published plugin. (#1778, #1774)\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-17T05:43:37Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v6.0.0\",\"name\":\"v6.0.0\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v6.0.0\",\"body\":\"## v6.0.0 (2026-06-16)\\n\\nSuperpowers 6.0 is a big release. The headline is a rewrite of how `subagent-driven-development` reviews each task \u2014 cheaper, stricter, and harder to game. \\n\\nWhile these numbers won't hold on every harness and for every workload, in our evals, Claude Code and Codex produce similar high-quality results roughly twice as fast and while spending almost 50% fewer tokens.\\n\\nIt also adds three new harnesses (Kimi Code, Pi, and Antigravity), gives the brainstorming visual companion a better security model, and rewrites a number of skills' tool calls to be significantly more vendor-neutral.\\n\\n### Visible Changes\\n\\n- **The two per-task reviewer prompts became one.** `spec-reviewer-prompt.md` and `code-quality-reviewer-prompt.md` are gone, replaced by a single `task-reviewer-prompt.md`. If you dispatch the old files directly, switch to the new one.\\n- **The legacy global worktree directory is gone.** `using-git-worktrees` and `finishing-a-development-branch` no longer use `~\/.config\/superpowers\/worktrees\/`. Worktrees now land in the project \u2014 an existing `.worktrees\/` or `worktrees\/` if you have one, otherwise a fresh `.worktrees\/` \u2014 unless you say otherwise.\\n\\n### New Harness Support\\n\\nSuperpowers now runs on three more harnesses. Each ships its own bootstrap, a tool-mapping reference, and tests, and each gets its own install section in the README.\\n\\n- **Kimi Code** \u2014 a plugin manifest, install docs, and manifest tests; install from Kimi's marketplace or straight from the repo. (initial manifest by @qer)\\n- **Pi** \u2014 a session-start extension that registers the skills and injects the `using-superpowers` bootstrap. Pi has native skills, so it needs no compatibility shim.\\n- **Antigravity (`agy`)** \u2014 installs the plugin directly and bootstraps from the first message; verified end-to-end against the standard \\\"make a react todo list\\\" acceptance test.\\n\\n### Subagent-Driven Development\\n\\nA long run of cost-and-quality experiments on real projects reshaped how the controller reviews each task. The old flow ran two reviewers per task and leaned on the controller's judgment for model choice and severity, and both turned out to be expensive and easy to game. The new flow runs one reviewer per task, hands work off as files instead of pasted text, and takes several judgment calls away from the controller.\\n\\n- **One reviewer per task, two verdicts.** A single `task-reviewer-prompt.md` reads the task's diff once and returns both a spec-compliance verdict and a quality verdict, so one fix pass clears both. A new \\\"can't verify from the diff\\\" verdict flags requirements that live in untouched code, for the controller to check itself. (#1538, #1543)\\n- **One broad review at the end.** The run finishes with a single whole-branch review on the most capable model, instead of re-reviewing everything task by task.\\n- **Plans get a pre-flight read.** Before the first task, the controller checks the plan for internal conflicts \u2014 and for anything the plan asks for that a reviewer would flag as a defect \u2014 and raises it all at once, rather than stumbling into it mid-run.\\n- **Diffs and task text move as files.** A pasted diff parks itself permanently in the most expensive context, and a reviewer without one rebuilds it by hand \u2014 the single biggest reviewer cost. Two new scripts, `task-brief` and `review-package`, write the task text and the review diff to files for the subagent to read.\\n- **Every dispatch states its model.** Left to choose, controllers stopped naming a model at all \u2014 and an unnamed model quietly inherits the session's most expensive one, so one run put all 26 of its reviewers on the top tier. The templates now require a model, with guidance that reaches for cheaper tiers when the work allows.\\n- **The controller can't tell a reviewer what to ignore.** Real runs caught controllers coaching reviewers to skip a finding or call it \\\"Minor at most,\\\" and the flaw shipped. Suppressing findings and pre-rating severity are now banned outright, and a defect the plan itself mandates gets reported for you to decide on rather than waved through.\\n- **Reviewers are read-only and skeptical of rationales.** Review no longer touches the working tree or branch \u2014 a reviewer running `git checkout` had been orphaning later commits \u2014 and an implementer's \\\"I left this unabstracted on purpose\\\" no longer talks a reviewer out of a real finding.\\n- **Stronger evidence and reporting.** Reviewers back each answer with a file and line, the implementer's report moves to a file and carries red\/green evidence when TDD applies, and a progress ledger lets a controller that loses its context resume instead of redoing finished work. (#994)\\n\\n### Writing Plans\\n\\nPlans now carry the structure the controller and reviewers used to re-derive on every dispatch.\\n\\n- **A Global Constraints block** lists the rules that bind every task \u2014 version floors, dependency limits, naming and copy, exact values \u2014 copied in verbatim, so they actually reach the implementers and reviewers downstream.\\n- **A per-task Interfaces block** names exactly what each task consumes and produces, so an implementer who sees only its own task still knows its neighbors' contracts.\\n- **Right-sizing guidance** keeps a task at the size that earns its own test cycle and a reviewer's pass, folding setup, config, and docs into the task that needs them. In testing, a plan written this way needed one round of fixes where the control needed two to four \u2014 and the control shipped a real bug.\\n\\n### Brainstorming Visual Companion\\n\\nThe visual companion is a small web server the agent opens alongside the conversation. It had no authentication at all, so on a shared or remote machine anyone who could reach the port could read your brainstorm \u2014 or inject events the agent treats as your input. This release gives it a real security model and makes it survive restarts and dropped connections.\\n\\n- **A per-session key now guards everything.** The agent's URL carries a one-time key, the browser tucks it into a tab-scoped cookie, and every request and WebSocket connection has to present it. This closes the door to stray local tabs and routable remote hosts alike, including the DNS-rebinding case an origin allowlist can't catch. (Closes #1014)\\n- **The file server stays in its sandbox.** It refuses symlinks, dotfiles, and any path that climbs out of the content directory, ignores macOS resource-fork files, and sends the usual no-store and deny-framing headers. Files that hold the session key are written owner-only.\\n- **The companion is offered only when it helps.** The skill raises it the first time a question would read better shown than told, as its own message, and lets a decline stand. Accepting opens your browser to the first screen. (Closes #755)\\n- **It survives restarts and flaky connections.** Given a project directory, the server keeps the same port and key across restarts, so an open tab simply reconnects. The page reconnects on its own, shows a live status pill, and raises a \\\"paused\\\" overlay while the server is down.\\n- **Longer idle life, safer shutdown.** The idle timeout went from 30 minutes to 4 hours, and `stop-server.sh` now confirms it owns the right process before signaling, so it never kills an unrelated `node` after a reboot. (#1703)\\n- **Windows launch hardening** \u2014 consolidated shell detection, and Windows now relies on the idle timeout for shutdown, since Node can't track POSIX process ownership across MSYS2.\\n\\n\\n### Existing Harness Updates\\n\\n- **Codex** now bootstraps through its own SessionStart hook rather than shared wiring, and the Codex App gained an install section and fuller tool docs (web search, `AGENTS.md`, personal skills). (#1540)\\n- **OpenCode** got an action-based tool mapping across its plugin, install doc, and README, plus a bootstrap-caching test.\\n- **Cursor**'s manifest dropped its `agents` and `commands` entries, since those directories no longer exist.\\n\\n### One Set of Skills, Every Harness\\n\\nThe skills used to speak Claude Code's dialect \u2014 \\\"use the Task tool,\\\" \\\"put it in CLAUDE.md.\\\" This release rewrites that vocabulary in terms of what you're actually doing (\\\"dispatch a subagent,\\\" \\\"your instructions file\\\") and adds a per-harness reference that maps each action to the right tool, checked against each runtime. Prose that named \\\"Claude\\\" now says \\\"your agent.\\\"\\n\\n- **A tool reference per harness** at `skills\/using-superpowers\/references\/`, covering Claude Code, Codex, Copilot, Gemini, Pi, and Antigravity.\\n- **`finishing-a-development-branch` went forge-neutral** \u2014 it no longer hardcodes `gh pr create`, so agents push with whatever forge tooling they have. (#1609)\\n- **One rename:** \\\"Claude Search Optimization\\\" is now \\\"Skill Discovery Optimization,\\\" since the technique isn't Claude-specific.\\n\\n### Writing Skills\\n\\nTwo additions for skill authors.\\n\\n- **Match the Form to the Failure** \u2014 a short table for picking the right kind of guidance. A flat \\\"don't do X\\\" works for discipline slips but backfires when the problem is the *shape* of an output, where a worked example does better. The table, and a tighter scope on the existing rationalization section, steer authors to the form that actually helps.\\n- **Micro-Test Wording** \u2014 a cheap way to check a phrasing before committing to it: sample it a handful of times against a no-guidance control and read every result by hand, treating run-to-run variance as a warning sign.\\n\\n### Testing\\n\\nSkill-behavior testing moved out of `tests\/` into a new `evals\/` submodule built on \\\"drill,\\\" which runs real Claude Code, Codex, and Gemini sessions and judges them with an LLM. Several in-tree bash suites retired once a stricter drill scenario covered them; the few with no equivalent stayed. From here on, `tests\/` holds plugin-code tests and `evals\/` holds skill-behavior tests, and `docs\/testing.md` explains the split. New backends reach Antigravity, Pi, and more models, and new shell-lint and pre-commit checks guard the harness. (#1541)\\n\\n### Bug Fixes\\n\\n- **systematic-debugging no longer forces every session into extended thinking.** One bullet held the exact keyword Claude Code scans for, quietly tripping the switch on every session that loaded the skill. A hyphen breaks the keyword; the text still reads. (#1283, by @Nick Galatis)\\n- **The Windows SessionStart hook stopped printing a write error every session** \u2014 each `printf` now routes through `cat` to absorb the broken pipe, and the output is otherwise unchanged. (#1612, reported by @silvertakana)\\n- **Windows foreground mode** tracks the right process and clears its owner PID on MSYS2. (by @nestorluiscamachopaz)\\n- **The `using-superpowers` bootstrap** no longer lists \\\"debugging\\\" as a skill that doesn't exist. (reported by @mhat)\\n- **The TDD skill** links the testing anti-patterns reference. (#1532, #1529; link fix #1474 by @Stable Genius)\\n- **`using-git-worktrees`** fixes its step numbering and drops stale Cursor references. (#1522, and by @fuleinist)\\n- **The Codex review skill** swaps a private in-joke for plain guidance. (#1531)\\n\\n### Documentation & Contributor Guidelines\\n\\n- **A guide to porting Superpowers to a new harness** (`docs\/porting-to-a-new-harness.md`) lays out the three pieces every integration needs and the one rule that makes or breaks it: load the bootstrap at session start.\\n- **Every PR and issue now discloses how it was made** \u2014 model, harness, version, and installed plugins, or a note that it was written by hand. We weigh a contribution differently depending on what produced it. PRs also target `dev`, not `main`. The PR template, all three issue templates, and a new platform-support template carry this.\\n\\n### Contributors\\n\\nThanks to @mattvanhorn, @nawfal, @Nick Galatis, @silvertakana, @nestorluiscamachopaz, @qer, @mhat, @Stable Genius, @fuleinist, @dev_Hakaze, @robotsnh, Rahul, and @arittr.\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-06-16T17:47:24Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]},{\"tagName\":\"v5.1.0\",\"name\":\"v5.1.0\",\"url\":\"https:\/\/github.com\/obra\/superpowers\/releases\/tag\/v5.1.0\",\"body\":\"### Removals\\n\\n- **Legacy slash commands removed** \u2014 `\/brainstorm`, `\/execute-plan`, and `\/write-plan` are gone. They were deprecated stubs that did nothing but tell the user to invoke the corresponding skill. Invoke `superpowers:brainstorming`, `superpowers:executing-plans`, and `superpowers:writing-plans` directly instead. (#1188)\\n- **`superpowers:code-reviewer` named agent removed** \u2014 the agent was the plugin's only named agent and was used by exactly two skills, while every other reviewer\/implementer subagent in the repo dispatches `general-purpose` with a prompt template alongside its skill. The agent's persona and checklist have been merged into `skills\/requesting-code-review\/code-reviewer.md` as a self-contained Task-dispatch template. Anyone dispatching `Task (superpowers:code-reviewer)` should switch to `Task (general-purpose)` with the prompt template instead. (PR #1299)\\n- **Integration sections removed from skills** \u2014 these were a legacy of the time before agents had native skills systems and didn't help with steering.\\n\\n### Worktree Skills Rewrite\\n\\n`using-git-worktrees` and `finishing-a-development-branch` now detect when the agent is already running inside an isolated worktree and prefer the harness's native worktree controls before falling back to `git worktree`. Behavior was TDD-validated and cross-platform-checked across five harnesses. (PRI-974, PR #1121)\\n\\n- **Environment detection** \u2014 both skills check `GIT_DIR != GIT_COMMON` before doing anything; if already in a linked worktree, creation is skipped entirely. A submodule guard prevents false detection.\\n- **Consent before creating worktrees** \u2014 `using-git-worktrees` no longer creates worktrees implicitly; the skill asks the user first. Fixes #991 (subagent-driven-development was auto-creating worktrees without consent).\\n- **Native tool preference (Step 1a)** \u2014 when the harness exposes its own worktree tool (e.g. Codex), the skill defers to it. The user's stated preference is respected when expressed.\\n- **Provenance-based cleanup** \u2014 `finishing-a-development-branch` only cleans up worktrees inside `.worktrees\/` (created by superpowers); anything outside is left alone. Fixes #940 (Option 2 was incorrectly cleaning up worktrees), #999 (merge-then-remove ordering), and #238 (`cd` to repo root before `git worktree remove`).\\n- **Detached HEAD handling** \u2014 the finishing menu collapses to two options when there is no branch to merge from.\\n- **Hardcoded `\/Users\/jesse` paths** in skill examples replaced with generic placeholders. (#858, PR #1122)\\n\\n### Contributor Guidelines for AI Agents\\n\\nTwo new sections at the top of `CLAUDE.md` (symlinked to `AGENTS.md`) speak directly to AI agents. An audit of the last 100 closed PRs against this repo showed a 94% rejection rate driven by AI-generated slop: agents that didn't read the PR template, opened duplicates, fabricated problem descriptions, or pushed fork- or domain-specific changes upstream.\\n\\n- **Pre-submission checklist** \u2014 read the PR template, search for existing PRs, verify a real problem exists, confirm the change belongs in core, and show the human partner the complete diff before submitting.\\n- **What we will not accept** \u2014 third-party dependencies, \\\"compliance\\\" rewrites of skill content, project-specific configuration, bulk PRs, speculative fixes, domain-specific skills, fork-specific changes, fabricated content, and bundled unrelated changes.\\n- **New harness PRs require a session transcript** \u2014 most past new-harness integrations copied skill files or wrapped with `npx skills` instead of loading the `using-superpowers` bootstrap at session start. The acceptance test (\\\"Let's make a react todo list\\\" must auto-trigger `brainstorming` in a clean session) and a complete transcript are now required.\\n\\n### Codex Plugin Mirror Tooling\\n\\nNew `sync-to-codex-plugin` script mirrors superpowers into the OpenAI Codex plugin marketplace as `prime-radiant-inc\/openai-codex-plugins`. Path\/user-agnostic so any team member can run it. (PR #1165)\\n\\n- Clones the fork fresh into a temp directory per run, regenerates overlays inline, and opens a PR; auto-detects upstream from the script's own location and preflights `rsync`\/`git`\/`gh auth`\/`python3`.\\n- `--bootstrap` flag for first-time setup; `EXCLUDES` patterns anchored to source root; `assets\/` excluded.\\n- Mirrors `CODE_OF_CONDUCT.md`; drops the `agents\/openai.yaml` overlay.\\n- Seeds `interface.defaultPrompt` in the mirrored `plugin.json`. (PR #1180 by @arittr)\\n- Codex plugin files are committed to the source repo so the sync script uses canonical versions; Codex marketplace metadata is preserved.\\n\\n### OpenCode\\n\\n- **Bootstrap content cached at module level** \u2014 `getBootstrapContent()` was calling `fs.existsSync` + `fs.readFileSync` + frontmatter regex on every agent step (the `experimental.chat.messages.transform` hook fires on every step in OpenCode's agent loop). Now read once, cached for the session lifetime, with a null sentinel for the missing-file case. 15 regression tests cover cache behavior, fs call counts, the injection guard, the missing-file sentinel, and cache reset. (Fixes #1202)\\n- **Integration tests modernized**.\\n- **Install caveats clarified** in the README.\\n\\n### Code Review Consolidation\\n\\n`requesting-code-review` is now self-contained: the persona, checklist, and dispatch template live in `skills\/requesting-code-review\/code-reviewer.md` and the skill dispatches `Task (general-purpose)` directly. (PR #1299)\\n\\n- **Single source of truth** \u2014 the persona\/checklist that previously lived in both `agents\/code-reviewer.md` and the skill's placeholder template (and drifted independently) is now one file.\\n- **`subagent-driven-development` follows suit** \u2014 its `code-quality-reviewer-prompt.md` now dispatches `Task (general-purpose)` instead of the named agent.\\n- **Behavioral test added** \u2014 `tests\/claude-code\/test-requesting-code-review.sh` plants real bugs (SQL injection, plaintext password handling, credential logging) into a tiny project and asserts the dispatched reviewer flags every planted issue at Critical\/Important severity and refuses to approve the diff.\\n- **Codex and Copilot workaround docs trimmed** \u2014 the \\\"Named agent dispatch\\\" sections in `references\/codex-tools.md` and `references\/copilot-tools.md` documented how to flatten a named agent into a generic dispatch. With no named agents shipping, the workaround is unnecessary; both sections were dropped.\\n\\n### Subagent-Driven Development\\n\\n- **No more pause every 3 tasks** \u2014 the \\\"review after each batch (3 tasks)\\\" cadence in `requesting-code-review` (originally for `executing-plans`) was leaking into `subagent-driven-development`. Replaced with \\\"each task or at natural checkpoints\\\" plus an explicit continuous-execution directive.\\n- **SDD integration test now runs its assertions** \u2014 three independent bugs caused the test to silently bail before printing any verification results: an unresolved `..` segment in the working-dir path, a `set -euo pipefail` interaction with `find | sort | head -1` (SIGPIPE on the producer killed the script), and a missing `--plugin-dir` on the `claude -p` invocation that caused the test to load the installed plugin instead of the working tree. All three fixed; six verification tests now actually run against a real end-to-end SDD run.\\n\\n### Cursor\\n\\n- **Windows SessionStart hook** routed through `run-hook.cmd` instead of invoking the extensionless `session-start` script directly. Fixes Windows opening the file in an editor instead of running it. Also removed an accidental UTF-8 BOM from `hooks-cursor.json`.\\n\\n### Gemini CLI\\n\\n- **Subagent dispatch mapping** \u2014 Gemini's `Task` dispatch now maps to `@agent-name` \/ `@generalist`, with parallel subagent dispatch documented for independent tasks.\\n\\n### Skills\\n\\n- **Terminology cleanups** across skill content.\\n\\n### Documentation & Install\\n\\n- **Factory Droid installation instructions** added to README.\\n- **Quickstart install links** in README. (PR #1293 by @arittr)\\n- **Codex plugin install guidance** updated. (PR #1288 by @arittr)\\n- **Codex `wait` mapping corrected** to `wait_agent` in the tools reference.\\n- **Install order reorganized**; Codex install instructions cleaned up.\\n- **Removed vestigial `CHANGELOG.md`** in favor of `RELEASE-NOTES.md` as the single source. (PR #1163 by @shaanmajid)\\n- **Discord invite link** fixed; release announcements link and a detailed Discord description added to the Community section.\\n\\n### Community\\n\\n- @shaanmajid \u2014 vestigial `CHANGELOG.md` removal (PR #1163)\\n- @arittr \u2014 README quickstart install links (#1293), Codex plugin install guidance (#1288), `sync-to-codex-plugin` `interface.defaultPrompt` seed (#1180)\",\"bodyExcerpt\":\"\",\"bodyTruncated\":false,\"publishedAt\":\"2026-05-04T22:13:34Z\",\"prerelease\":false,\"draft\":false,\"assetCount\":0,\"assets\":[]}]","ai_share_view_count":5345,"ai_share_bookmark_count":240,"ai_share_first_party_view_count":62,"ai_share_github_stars":273350,"ai_share_github_watchers":1029,"ai_share_github_forks":24453,"ai_share_github_open_issues":340,"ai_share_open_source":true,"ai_share_repo_url":"https:\/\/github.com\/obra\/superpowers","ai_share_documentation_url":"https:\/\/claude.com\/plugins\/superpowers","creator_id":"github-obra","creator_slug":"obra","creator_name":"obra","creator_type":"person","creator_status":"external","creator_intro":"obra maintains this GitHub repository.","creator_github_url":"https:\/\/github.com\/obra","creator_avatar_url":"https:\/\/avatars.githubusercontent.com\/u\/45416?v=4","ai_share_github_readme_path":"README.md","ai_share_github_readme_sha":"09a91c6d03cfa3d874c5c33b6f8cc3d0b2d89645","ai_share_github_readme_hash":"58b70d974287d9e57a5f54991eae92d9f490a323823dc3c8ef46c5ea9f19c43a","ai_share_github_readme_language":"en","ai_share_github_readme_format":"markdown","ai_share_github_description":"An agentic skills framework & software development methodology that works.","ai_share_one_sentence_definition":"An agentic skills framework & software development methodology that works.","ai_share_one_sentence_definition_en":"An agentic skills framework & software development methodology that works.","ai_share_github_readme_url":"https:\/\/github.com\/obra\/superpowers\/blob\/main\/README.md","ai_share_github_readme_raw_url":"https:\/\/raw.githubusercontent.com\/obra\/superpowers\/main\/README.md","ai_share_github_readme_markdown":"# Superpowers\n\nSuperpowers is a complete software development methodology for your coding agents, built on top of a set of composable skills and some initial instructions that make sure your agent uses them.\n\n## Table of Contents\n\n- [How it works](#how-it-works)\n- [Commercial Services](#commercial-services)\n- [Getting Started](#installation)\n  - [Claude Code](#claude-code)\n  - [Antigravity](#antigravity)\n  - [Codex App](#codex-app)\n  - [Codex CLI](#codex-cli)\n  - [Cursor](#cursor)\n  - [Devin CLI](#devin-cli)\n  - [Factory Droid](#factory-droid)\n  - [Gemini CLI](#gemini-cli)\n  - [GitHub Copilot CLI](#github-copilot-cli)\n  - [Grok Build CLI](#grok-build-cli)\n  - [Kimi Code](#kimi-code)\n  - [OpenCode](#opencode)\n  - [Pi](#pi)\n  - [Hermes Agent](#hermes-agent)\n- [The Basic Workflow](#the-basic-workflow)\n- [Community](#community)\n- [What's Inside](#whats-inside)\n- [Philosophy](#philosophy)\n- [Contributing](#contributing)\n- [Updating](#updating)\n- [License](#license)\n- [Visual companion telemetry](#visual-companion-telemetry)\n\n## How it works\n\nIt starts from the moment you fire up your coding agent. As soon as it sees that you're building something, it *doesn't* just jump into trying to write code. Instead, it steps back and asks you what you're really trying to do. \n\nOnce it's teased a spec out of the conversation, it shows it to you in chunks short enough to actually read and digest. \n\nAfter you've signed off on the design, your agent puts together an implementation plan that's clear enough for an enthusiastic junior engineer with poor taste, no judgement, no project context, and an aversion to testing to follow. It emphasizes true red\/green TDD, YAGNI (You Aren't Gonna Need It), and DRY. \n\nNext up, once you say \"go\", it launches a *subagent-driven-development* process, having agents work through each engineering task, inspecting and reviewing their work, and continuing forward. It's not uncommon for your agent to work autonomously for a couple hours at a time without deviating from the plan you put together.\n\nThere's a bunch more to it, but that's the core of the system. And because the skills trigger automatically, you don't need to do anything special. Your coding agent just has Superpowers.\n\n## Commercial Services\n\nIf you're using Superpowers in enterprise and could benefit from commercial support, additional tooling, or managed spending, please don't hesitate to drop us a line at sales@primeradiant.com.\n\n## Installation\n\nInstallation differs by harness. If you use more than one, install Superpowers separately for each one.\n\n### Claude Code\n\nSuperpowers is available via the [official Claude plugin marketplace](https:\/\/claude.com\/plugins\/superpowers)\n\n#### Official Marketplace\n\n- Install the plugin from Anthropic's official marketplace:\n\n  ```bash\n  \/plugin install superpowers@claude-plugins-official\n  ```\n\n#### Superpowers Marketplace\n\nThe Superpowers marketplace provides Superpowers and some other related plugins for Claude Code.\n\n- Register the marketplace:\n\n  ```bash\n  \/plugin marketplace add obra\/superpowers-marketplace\n  ```\n\n- Install the plugin from this marketplace:\n\n  ```bash\n  \/plugin install superpowers@superpowers-marketplace\n  ```\n\n### Antigravity\n\nInstall Superpowers as a plugin from this repository:\n\n```bash\nagy plugin install https:\/\/github.com\/obra\/superpowers\n```\n\nAntigravity runs the plugin's session-start hook, so Superpowers is active from\nthe first message. Reinstall with the same command to update.\n\n### Codex App\n\nSuperpowers is available via the [official Codex plugin marketplace](https:\/\/github.com\/openai\/plugins).\n\n- In the Codex app, click on Plugins in the sidebar.\n- You should see `Superpowers` in the Coding section.\n- Click the `+` next to Superpowers and follow the prompts.\n\n### Codex CLI\n\nSuperpowers is available via the [official Codex plugin marketplace](https:\/\/github.com\/openai\/plugins).\n\n- Open the plugin search interface:\n\n  ```bash\n  \/plugins\n  ```\n\n- Search for Superpowers:\n\n  ```bash\n  superpowers\n  ```\n\n- Select `Install Plugin`.\n\n### Cursor\n\n- In Cursor Agent chat, install from marketplace:\n\n  ```text\n  \/add-plugin superpowers\n  ```\n\n- Or search for \"superpowers\" in the plugin marketplace.\n\n### Devin CLI\n\n- Install the plugin from this repository:\n\n  ```bash\n  devin plugins install obra\/superpowers\n  ```\n\n- Update to the latest version with:\n\n  ```bash\n  devin plugins update superpowers\n  ```\n\n### Factory Droid\n\n- Register the marketplace:\n\n  ```bash\n  droid plugin marketplace add https:\/\/github.com\/obra\/superpowers\n  ```\n\n- Install the plugin:\n\n  ```bash\n  droid plugin install superpowers@superpowers\n  ```\n\n### Gemini CLI\n\n- Install the extension:\n\n  ```bash\n  gemini extensions install https:\/\/github.com\/obra\/superpowers\n  ```\n\n- Update later:\n\n  ```bash\n  gemini extensions update superpowers\n  ```\n\n### GitHub Copilot CLI\n\n- Register the marketplace:\n\n  ```bash\n  copilot plugin marketplace add obra\/superpowers-marketplace\n  ```\n\n- Install the plugin:\n\n  ```bash\n  copilot plugin install superpowers@superpowers-marketplace\n  ```\n\n### Grok Build CLI\n\nSuperpowers is available via the [official Grok plugin marketplace](https:\/\/github.com\/xai-org\/plugin-marketplace).\n\n- Install the plugin from xAI's official marketplace:\n\n  ```bash\n  grok plugin install superpowers@xai-official --trust\n  ```\n\n- Or open the marketplace in the TUI, search for Superpowers, and install it:\n\n  ```text\n  \/marketplace\n  ```\n\n### Kimi Code\n\nSuperpowers is available in Kimi Code's plugin marketplace.\n\n- Open Kimi Code's plugin manager:\n\n  ```text\n  \/plugins\n  ```\n\n- Go to `Marketplace` > `Superpowers` and install it.\n\n- Or install directly from this repository:\n\n  ```text\n  \/plugins install https:\/\/github.com\/obra\/superpowers\n  ```\n\n- Detailed docs: [docs\/README.kimi.md](docs\/README.kimi.md)\n\n### OpenCode\n\nOpenCode uses its own plugin install; install Superpowers separately even if you\nalready use it in another harness.\n\n- Tell OpenCode:\n\n  ```\n  Fetch and follow instructions from https:\/\/raw.githubusercontent.com\/obra\/superpowers\/refs\/heads\/main\/.opencode\/INSTALL.md\n  ```\n\n- Detailed docs: [docs\/README.opencode.md](docs\/README.opencode.md)\n\n### Pi\n\nInstall Superpowers as a Pi package from this repository:\n\n```bash\npi install git:github.com\/obra\/superpowers\n```\n\nFor local development, run Pi with this checkout loaded as a temporary package:\n\n```bash\npi -e \/path\/to\/superpowers\n```\n\nThe Pi package loads the Superpowers skills and a small extension that injects the `using-superpowers` bootstrap at session startup and again after compaction. Pi has native skills, so no compatibility `Skill` tool is required. Subagent and task-list tools remain optional Pi companion packages.\n\n### Hermes Agent\n\nInstall Superpowers as a Hermes plugin from this repository:\n\n```bash\nhermes plugins install obra\/superpowers --enable\n```\n\nRestart any active Hermes sessions after installing. Note: Hermes has no\npost-compaction hook, so a very long session that compacts over its first\nturn loses the bootstrap \u2014 start a fresh session if skills stop triggering.\n\n## The Basic Workflow\n\n1. **brainstorming** - Activates before writing code. Refines rough ideas through questions, explores alternatives, presents design in sections for validation. Saves design document.\n\n2. **using-git-worktrees** - Activates after design approval. Creates isolated workspace on new branch, runs project setup, verifies clean test baseline.\n\n3. **writing-plans** - Activates with approved design. Breaks work into bite-sized tasks (2-5 minutes each). Every task has exact file paths, complete code, verification steps.\n\n4. **subagent-driven-development** or **executing-plans** - Activates with plan. Dispatches fresh subagent per task with two-stage review (spec compliance, then code quality), or executes in batches with human checkpoints.\n\n5. **test-driven-development** - Activates during implementation. Enforces RED-GREEN-REFACTOR: write failing test, watch it fail, write minimal code, watch it pass, commit. Deletes code written before tests.\n\n6. **requesting-code-review** - Activates between tasks. Reviews against plan, reports issues by severity. Critical issues block progress.\n\n7. **finishing-a-development-branch** - Activates when tasks complete. Verifies tests, presents options (merge\/PR\/keep\/discard), cleans up worktree.\n\n**The agent checks for relevant skills before any task.** Mandatory workflows, not suggestions.\n\n## Community\n\nSuperpowers is built by [Jesse Vincent](https:\/\/blog.fsck.com) and the rest of the folks at [Prime Radiant](https:\/\/primeradiant.com).\n\n- **Discord**: [Join us](https:\/\/discord.gg\/35wsABTejz) for community support, questions, and sharing what you're building with Superpowers\n- **Issues**: https:\/\/github.com\/obra\/superpowers\/issues\n- **Release announcements**: [Sign up](https:\/\/primeradiant.com\/superpowers\/) to get notified about new versions\n\n## What's Inside\n\n### Skills Library\n\n**Testing**\n- **test-driven-development** - RED-GREEN-REFACTOR cycle (includes testing anti-patterns reference)\n\n**Debugging**\n- **systematic-debugging** - 4-phase root cause process (includes root-cause-tracing, defense-in-depth, condition-based-waiting techniques)\n- **verification-before-completion** - Ensure it's actually fixed\n\n**Collaboration** \n- **brainstorming** - Socratic design refinement\n- **writing-plans** - Detailed implementation plans\n- **executing-plans** - Batch execution with checkpoints\n- **dispatching-parallel-agents** - Concurrent subagent workflows\n- **requesting-code-review** - Pre-review checklist\n- **receiving-code-review** - Responding to feedback\n- **using-git-worktrees** - Parallel development branches\n- **finishing-a-development-branch** - Merge\/PR decision workflow\n- **subagent-driven-development** - Fast iteration with two-stage review (spec compliance, then code quality)\n\n**Meta**\n- **writing-skills** - Create new skills following best practices (includes testing methodology)\n- **using-superpowers** - Introduction to the skills system\n\n## Philosophy\n\n- **Test-Driven Development** - Write tests first, always\n- **Systematic over ad-hoc** - Process over guessing\n- **Complexity reduction** - Simplicity as primary goal\n- **Evidence over claims** - Verify before declaring success\n\nRead [the original release announcement](https:\/\/blog.fsck.com\/2025\/10\/09\/superpowers\/).\n\n## Contributing\n\nThe general contribution process for Superpowers is below. Keep in mind that we don't generally accept contributions of new skills and that any updates to skills must work across all of the coding agents we support.\n\n1. Fork the repository\n2. Switch to the 'dev' branch\n3. Create a branch for your work\n4. Follow the `writing-skills` skill for creating and testing new and modified skills\n5. Submit a PR, being sure to fill in the pull request template.\n\nSkill-behavior tests use the drill eval harness from [superpowers-evals](https:\/\/github.com\/prime-radiant-inc\/superpowers-evals\/), cloned into `evals\/` \u2014 see `evals\/README.md` for setup. Plugin-infrastructure tests live at `tests\/` and run via the relevant `run-*.sh` or `npm test`.\n\nSee `skills\/writing-skills\/SKILL.md` for the complete guide.\n\n## Updating\n\nSuperpowers updates are somewhat coding-agent dependent, but are often automatic.\n\n## License\n\nMIT License - see LICENSE file for details\n\n## Visual companion telemetry\n\nBecause skills and plugins don't provide any feedback to creators, we have no idea how many of you are using Superpowers. By default, the Prime Radiant logo on brainstorming's optional visual companion feature is loaded from our website. It includes the version of Superpowers in use. It does not include any details about your project, prompt, or coding agent. We don't see your clicks or anything about what you're building. This helps us have a rough idea of how many folks are using Superpowers and which version of Superpowers they're using. It's 100% optional. To disable this, set the environment variable `SUPERPOWERS_DISABLE_TELEMETRY` to any true value. Superpowers also honors Claude Code's `DISABLE_TELEMETRY` and `CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC` opt-outs."},"categories":[2462],"tags":[1911,1924,1927,1930,1933,1936,1939,1942],"ai_series":[],"ai_topic":[2650],"ai_creator":[716],"ai_collection":[15],"ai_resource_type":[105],"ai_tool_tag":[],"class_list":["post-4140","post","type-post","status-publish","format-standard","hentry","category-workflow_skill_pack","tag-ai","tag-brainstorming","tag-coding","tag-obra","tag-sdlc","tag-skills","tag-subagent-driven-development","tag-superpowers","ai_topic-programming-development","ai_creator-obra","ai_collection-ai-","ai_resource_type-workflow_skill_pack"],"source_url":"","preserve_modified":false,"lang":"en","translations":{"en":4140,"zh-cn":66},"pll_sync_post":{},"_links":{"self":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts\/4140","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/comments?post=4140"}],"version-history":[{"count":1,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts\/4140\/revisions"}],"predecessor-version":[{"id":4141,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/posts\/4140\/revisions\/4141"}],"wp:attachment":[{"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/media?parent=4140"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/categories?post=4140"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/tags?post=4140"},{"taxonomy":"ai_series","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_series?post=4140"},{"taxonomy":"ai_topic","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_topic?post=4140"},{"taxonomy":"ai_creator","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_creator?post=4140"},{"taxonomy":"ai_collection","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_collection?post=4140"},{"taxonomy":"ai_resource_type","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_resource_type?post=4140"},{"taxonomy":"ai_tool_tag","embeddable":true,"href":"https:\/\/ai.wuaishare.cn\/hub\/wp-json\/wp\/v2\/ai_tool_tag?post=4140"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}