My Agent Skills

Provides commands and patterns for Data Version Control (DVC). Use when the user asks about versioning large files, creating pipelines, or reproducing experiments.

Provides MLflow documentation covering Tracking, Projects, Models, and Registry. Use when the user asks about MLflow features, APIs, implementation, or debugging.

Performs static analysis, security scanning, and code quality auditing to detect vulnerabilities, secrets, and anti-patterns.

Systematic Root-Cause Analysis (RCA) and debugging protocol. Forces system awareness through `DEBUG_CONTEXT.md` and Mermaid diagrams before any code modification.

Builds transparent valuation ranges using scenario-based DCF and market multiple checks. Use when the user asks for intrinsic value, fair value range, margin of safety, or valuation sensitivity.

Design methodology for charts, dashboards, and data visualizations in any medium or library — form selection by data relationship, a disciplined color system, decluttering rules, and finance-specific conventions. Use before writing any chart code or when the user mentions 圖表, 視覺化, 儀表板, dashboard, plot, chart, or 報告配圖.

Use when receiving code review feedback to enforce engineering rigor. Ensures feedback is verified, understood, and technically evaluated before implementation, preventing performative agreement and blind changes.

Post-implementation cleanup pass that hunts reuse, simplification, efficiency, and abstraction-level (altitude) improvements in changed code while preserving behavior. Use after a feature works and is verified, or when the user says 重構, 精簡, 去重, 簡化, remove duplication, or clean up.

Systematic deep-investigation methodology for auditing codebases, evaluating systems, and researching complex questions. Enforces breadth-first mapping, claim-evidence discipline, cross-artifact consistency checks, and negative-space analysis. Use when the user asks for a deep review, audit, investigation, 深度檢視, 盤點, 稽核, or when a conclusion will drive an important decision.

分析目標學程的校友 5-10 年職涯路徑——職位、公司、薪資級距、地理分佈。用於驗證「這個學位真的能帶你到想去的地方」。

掃描目標市場的 AI Infra / MLOps 職缺趨勢——薪資帶、技能需求、企業類型、成長性。用於評估留學目的地的就業前景。

General framework for evaluating Machine Learning models across Computer Vision, NLP, and Classical ML. Provides data-driven insights and benchmarks performance against baselines.

Expert system for Ultralytics YOLO workflows (including YOLO26). Use when the user asks about training, validation, inference, dataset setup, model export, tracking, or performance tuning in Ultralytics.

Verifies that a code change actually does what it claims by exercising the affected flow end-to-end and observing real behavior — not just tests or typecheck. Use before committing nontrivial changes, or when the user says 驗證, 實測, 跑起來確認, "does it actually work", or claims completion without runtime evidence.

Expert system for generating Agent Skills compatible with any LLM agent (Copilot, Gemini, Codex, Claude, etc.). Enforces folder structure, YAML standards, and best practices.

管理留學申請的完整時間軸——從準備考試到拿到 offer 的全流程 checklist。用於確保不遺漏關鍵截止日期。

系統性研究目標學校的 AI/MLOps/AI Infra 碩士學程——課程結構、師資、研究方向、產學合作、校友出路。用於選校決策前的資訊收集。

Extracts structured facts from SEC filings and earnings documents. Use when the user mentions 10-K, 10-Q, 8-K, EDGAR, MD&A, risk factors, or filing-based financial analysis.

在執行任何投資分析或組合調整之前，主動上網搜尋當前宏觀經濟與地緣政治重大事件。這是一個「閘門技能」——確保所有下游分析都建立在最新現實之上，而非過時假設。

Acts as a QA Auditor and Red Team for new Agent Skills. Verifies structure, safety guardrails, and compliance with global-rules.

計算不同學校 × 地區組合的留學財務 ROI——包含學費、機會成本、稅後薪資、複利終值。用於量化「這筆教育投資到底划不划算」。

Infrastructure Architect and Guardian of Reproducibility. Balances "Docker-First" for complex apps with "Venv-Efficiency" for lightweight projects.

Use this skill when creating, rewriting, structuring, or reviewing technical documentation in Chinese. Suitable for 技術文件, 文件撰寫, 教學文件, 操作指南, 架構說明, API 文件, README 重構, and troubleshooting docs. It classifies docs with Diataxis, defines audience and single goal, and generates a skimmable Markdown structure.

追蹤並比較各國工作簽證政策——畢業後留任路徑、續簽門檻、永居條件。用於評估留學目的地的長期可行性。

Implements a systematic Incident Review & Skill Evolution loop. Analyzes project failures or debugging outcomes to identify systemic gaps and updates `skills/` to prevent recurrence.

Interprets macroeconomic regime shifts and links them to sector and valuation implications. Use when the user asks about inflation, rates, growth cycles, policy stance, or macro risk scenarios.

主動收集即時市場數據（股價、大宗商品、匯率、債券殖利率、波動率指標），確保所有估值與組合分析都基於最新數據而非過時快照。

Governance loop for this skill library. Defines the single-source-of-truth map, the validation workflow, the definition of done for adding or changing skills, the evolution-log convention, and the periodic drift audit. Use when adding, renaming, moving, or retiring a skill, when updating bundles or routing, or when the user asks 維護, 盤點, 一致性檢查, or library health.

Professional trunk-based development with branch-for-release workflow. Focuses on automation, environment isolation, and immutable releases via tags.

Reviews a diff or pending changes for correctness bugs using a findings-first discipline — hypothesize concrete failure scenarios, verify each finding against the code before reporting, rank by severity. Use when the user asks for code review, 審查程式碼, 找 bug, PR review, or before merging a branch.

Guides Specification-Driven Development (SDD) from ambiguous requests to clear, testable specs before coding. Use when the user asks for 規格驅動開發, spec-first workflow, 先寫規格再寫程式, or requests requirements/design/tasks documents with staged review gates for medium-to-large engineering changes.

Converts approved designs or detailed requirements into actionable, atomic implementation steps. Use when the user approves a design or asks "how to implement".

對投資組合執行情境式壓力測試，量化特定宏觀事件（戰爭升級、油價飆升、央行意外決議、科技泡沫破裂）對持倉的預估影響。將「定性擔憂」轉化為「量化損益估算」。

在數據收集（collecting-market-data, scanning-macro-news）與分析鏈之間的強制驗證閘門。執行三層數據品質檢核：多源交叉對標、統計異常偵測、LLM 邏輯與真實性審查。遵循「Garbage In, Garbage Out」原則——所有下游分析的品質上限取決於此技能。

Creates and maintains agent onboarding files (AGENTS.md, CLAUDE.md, GEMINI.md, copilot-instructions) for a repository — what belongs in them, what must stay out, and how to verify every documented command actually works. Use when the user says init, 初始化專案, onboarding, 建立 AGENTS.md, or when starting agent work in an undocumented repo.

Evaluates long-horizon business quality using durable economics, governance signals, and capital allocation behavior. Use when the user asks about moat, quality, or long-term compounding potential.

將分析結論轉化為具體的倉位規模（Position Sizing）與風控參數（停損/證偽點）。解決「看對但賺不到錢」的執行斷層——在分析鏈（report）與實際下單之間，提供數學化的資金分配與動態風控框架。

建立並維護滾動式催化劑日曆，追蹤所有可能觸發持倉標的「重新定價」的事件與時間點。解決「知道標的被低估但不知道何時反映」的時機問題，讓投資決策從「我覺得便宜」進化為「便宜且即將有催化劑確認」。

Translates financial research questions into an executable analysis plan. Use when the user asks how to analyze a company, macro theme, or investment thesis in a structured and auditable way.

Professional pre-release observability hardening. Implements structured logging, traceability (trace_id), and decision-reason logging using a strict Allowlist-first redaction policy. Use for logging/埋點/追蹤ID/可觀測性/個資遮罩/日誌安全需求.

評估地緣政治事件（戰爭、制裁、貿易衝突、供應鏈斷裂）對金融資產的傳導機制與影響路徑。當宏觀掃描偵測到地緣政治紅色警報時使用。

Audits financial analysis outputs for evidence quality, assumption integrity, and bias control. Use when the user asks to review, challenge, or validate a thesis before decision-making.

Acts as a Socratic Product Manager to explore, clarify, and define new application ideas. Use when the user has a vague idea, wants to "brainstorm", or asks to "design this app".

解讀市場隱含預期（估值、利率期貨、選擇權、信用利差），比對用戶觀點與市場共識的差距，找出「預期差」——這是超額報酬的真正來源。解決「只看基本面不看市場已定價多少」的盲點。

建構前瞻性情境推演，將「現在的宏觀現實」延伸為「未來 3-12 個月的多路徑演化」，並產出機率加權的最佳配置建議。與 stress-testing-portfolio（防守導向）互補，本技能聚焦於進攻性的戰略站位。

主動掃描新興投資主題與結構性趨勢，在用戶提出研究需求之前識別潛在機會。解決「只能被動研究已知題材」的盲點，讓分析鏈從「你問我答」進化為「我主動發現」。

Verifies financial conclusions against primary web sources such as official company investor-relations pages and regulatory filings. Use when the user asks to fact-check analysis outputs, validate assumptions, or review summaries before final decisions.

Performs post-mortem analysis on investment decisions to improve long-term process quality. Use when a thesis fails, underperforms expectations, or requires systematic process upgrades.

Standardizes raw financial statements into a consistent analysis schema. Use when the user asks for ratio analysis, trend comparison, or cross-company financial comparison.

對投資組合的績效進行多層歸因分析（Alpha/Beta 分離、技能層歸因、決策品質評分），並將結論回饋至上游技能以驅動系統自我進化。這是整條分析鏈的「閉環終端」——沒有它，系統只會重複犯同樣的錯誤。

Translates evidence-backed financial research into high-energy Traditional Chinese market storytelling with Xie-like cadence. Use when the user asks for podcast scripts, social posts, or quick commentary based on existing report facts and citations.

Defines and maintains long-term investment policy guardrails. Use when the user asks for portfolio discipline, rebalancing rules, risk limits, or decision frameworks for multi-year investing.

Creates new agent skills from a user-specified knowledge folder. Use when the user asks to convert notes, docs, templates, or domain references into reusable SKILL.md packages.

Translates evidence-backed financial research into structured Traditional Chinese narrative with Vincent-like long-form explanatory cadence. Use when the user asks for educational deep-dives, podcast scripts, or social posts that retain analytical rigor.