菜单
Base format for all report-style outputs (audits, triages, reviews). Defines item identification, severity sections, BLUF structure, and voice-friendly naming. Use when creating or modifying any skill that produces a report. Other report skills inherit from this.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Google Apps Script development best practices learned from production. Use when building Apps Script automation, payment API integrations, Google Sheets menu functions, modal dialogs, Poka Yoke quality gates, or running Apps Script functions from CLI. Covers UI dialog philosophy, API field discovery, submission architecture (source_amount vs transfer_amount), clasp CLI logging, run-appscript.sh function execution, and gate flow design.
Orchestration overview for the autonomous issue dispatch system. Handles bugs AND features from ANY entry point. References the component skills (dispatcher, handler) and per-project skills (bug-intake, qa-submission). Use when user says "dispatch", "fix these issues", "batch implement", "implement this", "build this", "develop this", "here's a plan", or any implementation/feature/fix request — including direct conversational requests to Claude Code.
Generalized Slack bug intake. Scans bug-report channels, auto-fixes small bugs, dispatches big work with clear specs, defers only genuinely ambiguous items to owner. Uses emoji reactions to track state. Each repo provides workspace config via local bug-intake skill. Supports shared channels where multiple repos receive bugs from one Slack channel via semantic routing.
PreToolUse hooks that block direct production database access and raw SQL write operations. Prevents bypass scripts that skip Drizzle ORM migration tracking, and HARD BLOCKS raw psql writes with no bypass — all writes must go through Admin API.
Standard development SOP for ALL code changes — bug fixes, features, refactors, any implementation. Defines the universal pipeline (investigate → fix → test → deploy → QA) and scope-based decision matrix. Also provides continuous scanning mode for autonomous operation. Use when implementing ANY code change, fixing bugs, building features, or when user says "start dev loop", "cowork", "fix this", "implement this", "build this". Each workspace can override with local skill.
Knowledge management for Claude Code projects. Use when: creating skills or agents, deciding where to put new knowledge (skill vs CLAUDE.md vs reference-data vs memory vs agent), routing information, CLAUDE.md hygiene, editing/modifying CLAUDE.md or skills or agents or memory files, extracting content from bloated files, or any discussion about knowledge architecture, documentation structure, or SOP placement. MUST be invoked BEFORE proposing any edits to CLAUDE.md, skills, agents, or memory - no exceptions.
| name | report-format |
| description | Base format for all report-style outputs (audits, triages, reviews). Defines item identification, severity sections, BLUF structure, and voice-friendly naming. Use when creating or modifying any skill that produces a report. Other report skills inherit from this. |
| user-invocable | false |
All report-style skills (audit-common, task-triage, employee-audit, etc.) inherit this format. Each consumer may override specific sections but must preserve the base principles.
Reports are discussed in voice and text. "About three" must never be ambiguous. Every item gets a globally unique identifier within the report. Numbers NEVER restart between sections.
Default: Adjective+Food names by severity category (STT-friendly):
| Severity | Names |
|---|---|
| 🔴 Critical/Urgent (red foods) | Spicy Tomato, Crispy Pepper, Blazing Chili, Red Cherry, Hot Salsa, Smoky Beet |
| 🟡 Warning/Important (yellow foods) | Golden Mango, Sunny Lemon, Tangy Orange, Warm Honey, Ripe Peach, Bright Corn |
| 🔵 Long Term/Info (blue foods) | Frosty Blueberry, Icy Grape, Misty Plum, Cool Fig, Deep Prune, Calm Olive |
| 🟢 Low Risk/Quick Win (green foods) | Fresh Apple, Minty Kiwi, Zesty Lime, Crisp Pear, Juicy Melon, Green Basil |
Override: Severity-prefixed sequential (for technical reports):
C1, C2, W3, W4, I5 — where prefix = severity, number = global sequence.
| Prefix | Severity |
|---|---|
C | 🔴 Critical |
W | 🟡 Warning |
I | ℹ️ Info |
Consumer declares which scheme it uses. Both are valid. Key rule: IDs are globally unique within the report.
Severity descending. Skip empty sections entirely (no "None." placeholders).
Every item follows this pattern. Fields can be renamed per domain but structure is mandatory:
**[ID]. [Title]** · [Source/Author]
[Platform icon]
> **Action:** [What to do — one sentence, BLUF. No context bleed.]
> **Context:** [Self-sufficient summary — reader understands without opening source]
> **Ref:** [clickable URL]
Rules:
| Platform | Icon |
|---|---|
| ✉️ | |
| GitHub | 🐙 |
| Google Tasks | ☑️ |
| Slack | 💬 |
| Audit finding | 🔍 |
| Schedule/Deferred | ⏰ |
End with **Actions?** — let the reader drive. Don't suggest batch actions.
Consumer skills declare: > **Inherits:** report-format at the top of their output format section. Then they state only their overrides:
> **Inherits:** report-format
**Overrides:**
- Item identification: uses C1/W3/I5 instead of food names
- Adds: Drift protocol section
- Adds: Registry logging requirement
If a consumer doesn't mention a base principle, it applies as-is.