Menu
SkillsMP a collecté 13 skills depuis MQ37/poisson. Ouvrez un skill pour examiner sa source et ses détails.
Affichage de 13 skills collectés sur 13.
Self-verification — spawn a fresh-context subagent to independently check that finished work actually satisfies the original request. Re-derives the task from a briefing you write (poisson subagents start with a blank conversation, not a copy of this one),…
Langue du texte source : anglais
Universal code-quality principles for any language or codebase — simplicity, clarity, and minimalism in the spirit of the suckless manifesto, Linus Torvalds' "good taste", a hacker's distrust of complexity, and the conviction that one mind can understand the…
Langue du texte source : anglais
General code review process — scan lenses (correctness, performance, edge cases, test coverage, security, API design, boundary/type cleanliness, orchestration/atomicity, ambitious simplification with a file-size ceiling and spaghetti-growth check, comment…
Langue du texte source : anglais
Convene a council of legendary engineers — spawned as parallel subagent personas — to review code, critique an architecture, or pressure-test an idea, then synthesize their verdicts into one ranked answer. Each legend carries a distinct, durable engineering…
Langue du texte source : anglais
Workflow for opening a pull request — picking a conventional-commit title, writing a tight What / Why / Testing description, and self-reviewing the diff before pushing. Language-agnostic and minimal in the spirit of suckless and Torvalds' "show me the code".…
Langue du texte source : anglais
Scout the blast radius of a change before or after writing it — name the shared seam it touches, enumerate from the code every mode, flag, header, and consumer already flowing through that seam, then classify each as works-unchanged / wired /…
Langue du texte source : anglais
Set up an isolated podman sandbox for a work session — mount the right directory (workdir, or a fresh git worktree when parallel branches could conflict) from the start, do all bash/build/test work through the container, edit files with the normal host-path…
Langue du texte source : anglais
Author a new SKILL.md — as a builtin shipped in the px binary (internal/skills/builtin/<name>/SKILL.md, embedded via go:embed, no other code change needed) or as a user skill under ~/.poisson/skills/<name>/SKILL.md (picked up live, /reload, overrides a…
Langue du texte source : anglais
Relentless one-question-at-a-time interview to stress-test a plan, design, or decision before acting on it — maps it as a design tree, works the open frontier each round, dispatches subagents for anything discoverable instead of asking the user. Use when the…
Langue du texte source : anglais
Red-green-refactor discipline — write a failing test before any production code, watch it fail for the right reason, write the minimum code to pass, then refactor with the suite green. Use when implementing new behavior, fixing a bug (reproduce it as a…
Langue du texte source : anglais
Workflow for drafting a tight, free-form issue — short, simple, focused on one thing. For bugs, captures what's broken and how to reproduce it (or where it was observed when it can't be reproduced). Use when the user asks to file, open, write, or draft an…
Langue du texte source : anglais
Gather a PR or branch diff for review — decision tree for local working tree vs GitHub PR vs a fresh /tmp checkout, dumped to a file so nothing is lost to terminal truncation. Use when the user asks to review a PR, summarize a diff, or audit changes; hands…
Langue du texte source : anglais
Output format for presenting a code review as a stacked, risk-tiered summary — every file classified 🔴 critical / 🟡 notable / 🟢 routine, critical-first ordering, OLD/NEW shape diffs, numbered human explanations with explicit risks, and a leading TL;DR. Use…
Langue du texte source : anglais