DennisWei9898

品牌無關的 IG/Threads/X/FB 圖卡引擎——把任何主題（idea / 文章 / 教學）變成輪播圖卡 + 4 平台文案。HTML+CSS 模板 → Chrome headless 渲染 PNG。風格由 brand pack 決定（brands/），接哪個 pack 出哪個風格，引擎不綁死品牌。觸發詞：做圖卡、生 carousel、social card、IG 圖卡、Threads 卡片、貼文文案、教學圖卡、infographic、輪播圖、carousel post。

知識型 IG 輪播圖卡的獨立審核員（Joker）——預設有罪推定、禁止 LGTM。對一套輪播圖卡＋文案，按「封面/結構/擴散/CTA/品牌」五類硬規則逐條 PASS/FAIL，指到卡給可照做的改法，並算社群擴散體質分。是 Social Post Producer 迴圈裡的「審」（寫審分離）。觸發詞：驗收這套輪播、這組圖卡會不會擴散、red team 這個 carousel、carousel joker、圖卡審核、這套 IG 圖文夠不夠好。立場是找碴不是稱讚。

迷因知識輪播的獨立審核員——用「幽默感」的可二元判定要素（不協調-解消/良性冒犯/切身共鳴/誇張反差/自嘲/三的法則/秒懂）＋迷因 Hook＋版權安全，對一套迷因圖卡逐條 PASS/FAIL、指到卡給改法。預設有罪推定、禁 LGTM。是迷因輪播 loop 的「審」。觸發：這張夠不夠好笑、驗收迷因圖、meme joker、這個梗會不會冷、迷因輪播審核。

Bridges paper-mentor with NotebookLM. Creates notebooks, adds PDF/URL sources, runs queries with citation-grounded answers. Used by `paper-mentor` Phase 2 to build a literature library and run the 4-question summary on each paper. Triggers on phrases involving NotebookLM operations + chemistry/materials context.

A virtual research advisor that walks graduate students through the entire research-to-writing process — picking a topic, finding papers, drafting, and final grading. Uses Socratic questioning instead of writing for the student. Loads a department-specific advisor profile (v0.2 ships with 4 profiles: chieh-ting-lin for chemistry/materials, social-science-default for sociology/polisci/psych/econ, humanities-default for literature/history/philosophy, cs-default for computer science/ML/systems). Triggers on phrases like 我想找教授討論、I want to talk to a professor、help me write my paper、幫我審論文、score my paper、研究方向、論文評分.

Searches Semantic Scholar and the open web for chemistry / chemical engineering / materials science papers given a topic and keyword sets. Returns a triaged table with title, first author, year, citation count, PDF availability, and one-line summary. Used as a sub-skill by `paper-mentor` during Phase 2; can also be invoked directly. Triggers on phrases like "find papers on", "幫我找文獻", "search Semantic Scholar", "literature on X".

Safely find, vet, and install any Claude Code skill through a 3-stage review pipeline. Trigger on "install skill", "find skill", "add skill", "安裝 skill", "找 skill", "幫我裝 skill", "新增 skill", "validate skill I made", "幫我驗證這個 skill", "審查我寫的 skill".

Audit any Claude Code skill for behavioral control and security issues before installation. Checks CLAUDE.md injection, tool hijacking, auto-commit, proactive workflow takeover, telemetry, and external promotions. Trigger on "audit skill", "check skill security", "is this skill safe", "安全審查 skill", "驗證 skill 安全性", "這個 skill 安不安全", "skill 有沒有資安問題".

Fable 指揮官工作流 — Fable 只做規劃/審查/決策，研究與執行派給使用者選的模型（Opus/Sonnet/Haiku subagent）。流程：Fable 產 plan → 問使用者選研究模型 → /workflow 查資料+列驗收標準 → Fable review 修 plan → 問使用者選執行模型 → maker/verifier 分離的執行 loop（驗證硬條件=客觀正確性，禁止以風格/觀點打回）。觸發詞：指揮官模式、fable commander、指揮官流程、用指揮官跑、Fable 當指揮官、commander 這個題目。

A reusable, tool-agnostic method for turning scattered product/operations documents, design files, and tutorial videos into a clean, citation-grounded, queryable knowledge base — built once as Single-Source-of-Truth Markdown and served through two outputs (a no-code Q&A tool such as NotebookLM + a reusable Claude Skill). Every answer can point back to the exact source it came from. Use when someone wants to build an internal "ask-anything" assistant for ops/support/PM teams, consolidate messy docs into a sourced knowledge base, set up a low-maintenance monthly-updated KB, resolve cross-file contradictions, run an adversarial integrity audit, ground answers in a real database schema, or wants the QA / anti-hallucination workflow for a document Q&A system.

Loop Engineering 合規審查員 — 拿 Addy Osmani《Loop Engineering》當量尺，審查一個 agent / skill / 自動化迴圈專案「產出/驗證分離」的設計對不對，逐條給 PASS/FAIL/WARN + 證據，然後在**保留使用者原始風格**的前提下只修合規問題。提問走八二法則（規則不清時先挑 3-5 個影響最大的問，不一次拋 50 題），循序漸進把合規度拉到 100%。觸發詞：loop engineering 審查、審一下這個 agent、這個 skill 合不合規、loop 合規檢查、review 這個迴圈、loop-engineering-reviewer、幫我體檢這個 agent。