Rigorous engineering mode for nontrivial work in this repo — a set of named principles plus the leaf skills that apply them. Use when the user says "pstack", "go deep", "be rigorous", "认真做", or when a task involves architecture, a real bug, a refactor, or anything the user will not be watching. Ported from cursor/plugins pstack by Lauren Tan (MIT); see ATTRIBUTION.md.
インストール方法を見るインストールする前に、エージェントに与えられる指示の中身を確認できます。
Rigor over throughput. The point is to write less code, and to be able to prove the code you wrote works.
Original by Lauren Tan (poteto), MIT, from the Cursor plugin marketplace. This is a port: the principles and general skills are upstream's, the orchestration around them is rewritten for Claude Code and this repo. What changed and why: ATTRIBUTION.md.
Read the principles below. Name the ones that shaped a decision, and say which choice each one changed. A citation with no decision behind it means you did not actually apply it — that is worse than not citing, because it reads as rigor without being rigor.
For any principle you actually apply, read its leaf skill in full
(skills/principle-<name>/SKILL.md). The one-liners here are an index, not
the content.
Core
principle-laziness-protocol). Refactoring, sizing a diff, or tempted to add abstractions, layers, or signal threading. Bias to deletion and the smallest change that solves the problem.principle-foundational-thinking). Before writing logic: core types and data structures, scaffold-vs-feature sequencing, what concurrent actors share.principle-redesign-from-first-principles). Integrating a new requirement into an existing design. Redesign as if it had been foundational from day one.principle-subtract-before-you-add). Sequencing an addition, refactor, or rewrite. Remove dead weight first, then build on the simpler base.principle-minimize-reader-load). Reviewing or shaping code that's hard to trace. Count layers and hidden state, collapse one-caller wrappers, shrink mutable scope.principle-outcome-oriented-execution). Planned rewrites and migrations with explicit phase boundaries. Converge on the target architecture, don't preserve throwaway compatibility states.principle-experience-first). Product, UX, or feature-scope tradeoffs. Choose user delight over implementation convenience.principle-exhaust-the-design-space). A novel interaction or architectural decision with no precedent. Build 2-3 competing prototypes and compare before committing.principle-build-the-lever). Any non-trivial work. Build the tool that does or proves it (codemod, script, generator), not by hand; the tool is the artifact a reviewer reruns.Architecture
principle-model-the-domain). Writing stateful logic, or code that branches a lot or repeats a shape assumption across files. Encode the domain in a structure instead of scattered conditionals.principle-boundary-discipline). Wiring validation, error handling, or framework adapters. Guards at system boundaries, trust internal types, keep business logic pure.principle-type-system-discipline). Designing types or a signature. Make illegal states unrepresentable, brand primitives, parse external data at boundaries.principle-make-operations-idempotent). Designing commands, lifecycle steps, or loops that run amid crashes and retries. Converge to the same end state.principle-migrate-callers-then-delete-legacy-apis). Introducing a new internal API while old callers exist. Migrate and delete in one wave.principle-separate-before-serializing-shared-state). Concurrent actors might write the same file, branch, key, or object. Eliminate the sharing first.Verification
principle-prove-it-works). After a task, before declaring done. Verify against the real artifact, not a proxy or "it compiles".principle-fix-root-causes). Debugging. Trace each symptom to its root cause, reproduce first, ask why until you reach it.principle-sequence-verifiable-units). Multi-step work and how you stack commits and PRs. Break work into units that each end in a check.Delegation
principle-guard-the-context-window). Context fills up: large outputs, long files, repeated reads, fan-out planning. Route bulk to subagents, keep summaries in the main thread.principle-never-block-on-the-human). Tempted to ask "should I do X?" on reversible work. Proceed, present the result, let the human course-correct.Meta
principle-encode-lessons-in-structure). You catch yourself writing the same instruction a second time. Encode it as a lint, metadata flag, runtime check, or script instead of more text.how.why. It fans out across this repo's decision record (ADRs, docs/design/, the issue store, the changelog, wisp) and git history in parallel, and cites or names the gap. how answers what the code does; why answers what forced its shape.architect (parallel design exploration before implementing).swarm. Design or code bakeoff → arena.interrogate (adversarial multi-reviewer) before shipping.unslop. Docs, RFCs, PR descriptions, commit messages → technical-writing.no-comments.show-me-your-work.AGENTS.md (symlinked as CLAUDE.md) is authoritative. Where pstack and it
disagree, the repo wins. The conflicts that actually come up:
main and new or moved keybindings need the owner's word in that turn
regardless of what this principle says.bun run lint, bun run test:fast (vitest) or bun test test/render (bun's
runner — picking the wrong one looks like a broken environment), and for
visual work the browser /harness path, never a local screenshot.Reversible work proceeds without asking. Always pause for irreversible writes AGENTS.md does not already authorize: force-push to shared branches, deploys, data deletion, messages to people.
Upstream's Graphite/bugbot workflows (shipping, babysit, autopilot-*,
orchestrate) were not ported — this repo uses plain gh PRs, and a playbook
that reaches for tools you do not have is worse than no playbook. The /pr
skill covers that ground. Details in ATTRIBUTION.md.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Sketch types, signatures, and module structure before code, then stay in the loop while implementation fills in. Use for /architect, 'architect this', 'design this', or non-trivial work where jumping to code would lock in the wrong shape.
日本語の概要は準備中です。原文の説明を表示しています。
Spawn N parallel candidates at the same task, pick a base, graft the strongest parts of the losers into it. Use for /arena, 'arena this', 'throw it in the arena', or when one attempt at a non-trivial artifact would lock in the wrong shape.
日本語の概要は準備中です。原文の説明を表示しています。
在 rove 仓库内跑 auto-motion——把 transcription.srt 拆成多段 MG 动画镜头并拼接成竖屏视频(storyboard 分镜 + theme.md 全片主题 + 逐镜头 claude -p 子进程 + ffmpeg 拼接)。本 skill 是薄 wrapper:解析 auto-motion 模板根,继承 rove 品牌 theme,执行逻辑以 auto-motion 仓库的 canonical SKILL.md 为准。当用户说"跑 auto-motion"、"把这个字幕稿/口播稿做成视频"、"给 rove 做一条 MG 宣传片"时使用。
日本語の概要は準備中です。原文の説明を表示しています。
Find what a change could break somewhere else before it ships, beyond the diff, and prove the one fact it's safe because of by running real code instead of writing it up. Use for 'blast radius of X', 'what could this break', or reviewing a small diff you don't trust.
日本語の概要は準備中です。原文の説明を表示しています。
Draft Rove release notes as Changesets. Writes user-facing entries as `.changeset/*.md` files for `@sma1lboy/rove` (consumed into `packages/rove/CHANGELOG.md` at release time). Use when the user asks for "changelog", "release notes", "what changed", "add a changeset", or before cutting a version. Enforces Rove's no-soft-wrap rule so GitHub release pages render flowing text.
日本語の概要は準備中です。原文の説明を表示しています。
Design an auditable playbook when no narrower one fits: a large migration, an ambitious multi-part change, or work a human reviews after stepping away. Scales rigor to the task, runs a hypothesis loop, and logs decisions via show-me-your-work. Use for /figure-it-out, 'figure it out', a large migration, or when no narrower playbook applies.
日本語の概要は準備中です。原文の説明を表示しています。