Feishu Doc Writing

Organizes and formats group discussions, meetings, and research into clear, complete Feishu documents with proper structure, visuals, and security.

Security Scan

The skill's writing rules align with its stated purpose, but it implies Feishu integration and ongoing learning from user edits without declaring any credentials, access mechanisms, or data-retention behavior — this mismatch deserves clarification before install.

Purpose & Capability

The name and description (Feishu doc writing) match the SKILL.md content (documenting meetings, structuring, visuals, permissions). However, the spec never declares how the skill will access or publish Feishu documents (no Feishu API keys, tokens, or connector instructions). That omission is notable but may be intentional if the platform provides built-in Feishu access — the skill should state that dependency.

Instruction Scope

Runtime instructions are limited to writing style, structure, signing, permission rules, and a security guideline to avoid leaking secrets. They do not instruct reading any local files or env vars. One instruction—"从用户后续修改中学习" (learn from user subsequent edits)—implies persistent learning or tracking of user edits but gives no detail on what is stored, where, or for how long; this needs clarification.

Install Mechanism

Instruction-only skill with no install steps and no code files, so there is nothing being downloaded or written to disk by the skill itself.

Credentials

No environment variables, credentials, or config paths are requested. Given the skill's explicit Feishu focus and permission rules (boss: full_access, members: view, requester: edit), it's reasonable to expect some Feishu credentials or platform connector requirements; their absence is a potential gap that should be explained.

Persistence & Privilege

always is false and autonomous invocation is allowed (default). The only persistence-like behavior is the guideline to learn from user edits; the skill does not declare if or how it will persist user edits or learning data. Confirm whether learning is local ephemeral behavior, platform-managed personalization, or stored elsewhere.

What to consider before installing

This skill is mostly a style/formatting guide for creating Feishu documents and is instruction-only (no code to run). Before installing: 1) Ask how the agent will access and publish Feishu docs — does the platform supply Feishu connectivity or will you need to provide API credentials? 2) Clarify the "learn from user edits" behavior: where are edits stored, how long is data retained, and is it shared with other agents? 3) Confirm enforcement of the stated security rule (never expose API keys/internal addresses) — ensure the platform or skill runtime will not leak source chat contents to external endpoints. 4) Test the skill with non-sensitive example data to verify permission mappings (boss/edit/view) and the author signature behavior. If these questions are answered satisfactorily (especially about Feishu integration and data retention), the skill appears coherent; if not, treat it as a potential privacy risk.

飞书文档写作 Skill

来源：黄哥在"养虾"中总结的写文档规范，经主人确认纳入。

触发条件

当需要将群聊讨论、会议内容、调研结果等整理为飞书文档时激活。

---

写作规范

1. 内容完整性

• 文档内容 ≥ 群聊/源素材内容，不能丢信息
• 主动补充背景、上下文、相关细节，让没参与讨论的人也能看懂

2. 层级美观

• 标题层级清晰（H1 → H2 → H3），不跳级
• 段落之间有"呼吸感"，不挤在一起
• 善用分隔线、空行，让结构一目了然

3. 图文结合

• 不堆文字墙，适当插入图片、表格、流程图
• 有数据用表格，有流程用图示
• 截图、示意图能让文档更直观

4. 加粗策略

• 重要内容、关键结论、核心数据加粗
• 从用户后续修改中学习：用户加粗了什么、去掉了什么加粗 → 调整自己的加粗判断
• 不滥用加粗，加粗太多 = 没加粗

5. 作者署名

• 格式：小龙虾🦞 × 提需求人
• 小龙虾🦞 是 AI 助手的统一署名
• 提需求人是发起文档需求的人

6. 权限规则

• 老板：full_access（完全访问）
• 群成员：view（只读）
• 提需求人：edit（可编辑）

7. 安全规则

• 不暴露内部敏感信息（API key、内部系统地址、未公开数据等）
• 对外文档额外注意脱敏