Menu
Replace with a description of the skill and when Kian should use it.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
飞书 Todo 后台同步 Skill。安装完成后，用户请在一个新对话里说 "启用 feishu-task-sync" 或 "install feishu-task-sync" 以触发安装。 Agent 必须立刻按本 SKILL.md 顶部的"激活规则"驱动完整安装流程：先让 用户用 permissions/required-scopes.json 批量导入飞书后台权限并发布 版本 → 收 config 字段 → 调 bootstrap.py install 走 OAuth → 自动 first-run 心跳 → 写 cron 并绑定专用后台 Agent。日常运行时本 Skill 每小时让 Agent 自己阅读最近飞书聊天/文档/Wiki 中 @用户 的内容并语义 提炼 Todo，调用 feishu_tasks.py 创建飞书任务并加用户为 assignee； 同时每小时心跳 + 每日 11:00 摘要走广播渠道，绝不污染主对话。
End-to-end research methodology skill. Use when the user asks Codex to plan, run, analyze, compare, review, or document research, experiments, benchmarks, ablations, empirical investigations, small-scale studies, or exploratory technical work. Covers question framing, hypothesis and variable design, fair experiment comparison, evidence analysis, report writing, and method distillation.
Load task-relevant context from the user's zz-wiki personal knowledge base. Use when a task mentions zz-wiki, cross-project memory, user preferences, prior decisions, project state, skill routing, or when Codex needs durable personal context before acting. Reads selectively instead of loading the entire wiki, and can check whether referenced skills from stupidZZ/skills are locally installed.
| name | template-skill |
| description | Replace with a description of the skill and when Kian should use it. |
Use this file as a starting point. The minimum required frontmatter is:
name: lowercase identifier, hyphen-separated.description: one or two sentences describing when Kian should activate the
Skill.Optional frontmatter that this repository recommends:
version: SemVer string for the Skill itself.homepage: link to documentation or repository for the Skill.tags: array of free-form labels.The body of SKILL.md is the runbook Kian follows when the Skill is active.
Keep instructions concrete and reference any sibling files (prompts/,
scripts/, …) with relative paths.