Turn raw client conversations, meeting transcripts, delivery notes, feedback, and business records into an evidence-backed business memory for solo service providers and small professional teams. Use when Codex needs to organize real operating material, extract verifiable facts and business events, update a client…
Use when building a curated video course website from YouTube — picking and verifying videos, organising them into chapters/units, attaching evidence or references, and shipping a static site with good SEO. Topic-agnostic; works for anatomy, cooking, guitar, statistics, welding, anything with good YouTube coverage.
Evidence-first public-sector research with official-source planning, bounded collection, passage-level citations, explainable scoring, local SQLite memory, and readable Korean Markdown plus interactive HTML reports. Use when an agent must investigate laws, government policy, public services, procurement, privacy…
Stage a live debate between your persona's antipattern and signature voices on a topic. Triggers: "crush", "debate personas", "personas fight", "/honne:crush".
Conflict synthesis from antipattern × signature axes. Generates a working persona prompt from your observed patterns. Triggers: "persona", "activate persona", "honne persona".
Use when the user asks about strength training, building a training program, hypertrophy, fat loss, gaining mass, nutrition plans, protein/macros, supplements (creatine, BCAA), recovery, cardio + lifting, training at home, or choosing a coach — any strength training or nutrition question. Triggers on "I want to lose…
Run The Crucible (圆桌萃鉴), a rigorous evidence-backed multi-reviewer review. Configure a confirmed table, preserve separate first passes, mechanically compare per-item findings, adjudicate objections against primary evidence, and produce traceable consensus and unresolved decisions. Use when the user explicitly asks for…
Invoke only when the user explicitly mentions $crux. When invoked, analyze papers, coach research, or support consequential decisions by finding the variable that could change the outcome, separating evidence from rhetoric, and controlling how much help to reveal. Do not activate for ordinary paper, research, or…
A work method for non-trivial tasks that produce conclusions or deliverables, such as research, analysis, troubleshooting, plans, documents, or code. It emphasizes checking evidence, especially when sources conflict or facts and numbers matter.
Free local academic paraphrase and evidence-safe writing skill. Use for meaning-preserving paraphrase, Indonesian-friendly academic English, research framing, literature writing, theses, methods/results/discussion, reviewer responses, evidential verbs, claim control, and original pattern retrieval. Do not use to…
Review user-facing software changes for accessibility barriers across semantics, keyboard and other input, focus, structure, content, feedback, and assistive technology behavior. Use for web, mobile, desktop, CLI, and interactive UI changes; adapt the checks to the platform.
Protect AI coding agents from instructions hidden in repository files, issues, documents, web pages, tool output, model context, or memory. Use whenever an agent reads untrusted content, retrieves external context, summarizes data, or proposes an action based on mixed instructions and data.
Review a proposed or existing SKILL.md before publication for practical scope, precise triggers, evidence quality, counterexamples, safety, overlap, composability, and honest lifecycle status. Use when creating, revising, promoting, deprecating, or merging agent skills.