personal-agent-skills

清理 GitHub 仓库的 star-farming bot 账号——拉取 stargazers、识别 bot、批量 block。仅在用户显式调用时使用：`/github-bot-cleanup <org/repo>` 或明确提及 'github bot 清理'、'清除虚假 star'、'star 农场'、'block bot 账号'。日常对话不要主动加载。

价值投资深度分析框架。当用户请求对个股进行基本面分析、护城河评估、DCF/相对估值、财务质量诊断、商业模式拆解、管理层评估、安全边际测算、买入卖出决策、行业对比、价值陷阱识别、深度研报时使用。覆盖 A股（.SH/.SZ/.BJ）、港股（.HK）、美股（NYSE/NASDAQ）。触发词：价值投资、个股分析、基本面、护城河、moat、DCF、自由现金流、ROIC、安全边际、内在价值、巴菲特、芒格、能力圈、价值陷阱、深度研报、买入逻辑、卖出逻辑、估值。Value investing deep-dive framework — triggers on fundamental analysis, moat assessment, DCF/relative valuation, financial quality diagnosis, business model decomposition, management evaluation, margin of safety, buy/sell decision, peer comparison, value-trap detection. Covers A-shares, HK, US equities. Keywords include value investing, stock analysis, fundamentals, moat, DCF, FCF, ROIC, intrinsic value, margin of safety, Buffett, Munger, circle of competence, value trap.

Generate a high-signal daily X intelligence report for OpenClaw. Use when producing recurring OpenClaw X monitoring briefs, KOL/watchlist updates, Chinese/English hot-post summaries, or action-oriented social intelligence outputs. Enforce strict quality gates (relevance, engagement, depth, structure) and reject low-quality filler posts.

Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.

Distill article links into stable Obsidian reading notes. Use when a user sends one or more article URLs, especially `mp.weixin.qq.com`, and expects summary, first-principles analysis, cleaned original text archiving, weekly-review candidate tracking, or durable reading-note ingestion. Also use when standardizing or repairing reading-note format, tags, filenames, or section structure.

Fetch 90 top Hacker News blogs (curated by Karpathy), score/categorize/summarize articles using the agent's own model, then deliver a chat-first daily digest, save a full Markdown report locally, and sync it to Obsidian. No external API key needed. Use when user mentions 'daily digest', 'RSS digest', 'blog digest', 'tech news', 'tech blog summary', or runs /ai-daily-digest. Trigger command: /ai-daily-digest.

Plan and execute deep research tasks with explicit scoping, phased search, source-tiering, multi-agent decomposition, red-team verification, and decision-ready synthesis. Use when the user asks for detailed research, broad/deep investigation, multi-round search, competitor/landscape analysis, technical deep dives, or wants a plan before execution and may benefit from subagents.

Plan and execute deep research with explicit scoping, light-search-first planning, source-tiering, role-based parallel decomposition, red-team verification, and decision-ready synthesis. Use when the user asks for deep research, broad investigation, multi-round search, landscape mapping, competitor analysis, technical diligence, executive briefs, recommendation support, plan-first research, or work that should benefit from subagents instead of a one-shot answer.

Create, improve, evaluate OpenClaw-compatible skills with structured iteration loops, eval design, review artifacts, and packaging guidance. Use when building a new skill, improving an existing skill, designing evals, comparing skill revisions, or tightening trigger descriptions and release readiness.

Transform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autonomous Crons, and battle-tested patterns. Part of the Hal Stack 🦞