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。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/CleanExpo/Unite-Group --skill delegation-planner命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
The authoritative environment-variable registry for the Unite-Group Nexus. Use whenever adding, reading, renaming, or debugging an env var or secret — new integration config, a `process.env.X` read, an auth/OAuth wiring, a cron secret, a "which variable holds this" question, or a "works locally but not in prod" symptom. Also use before pinning any variable name in code, and when reconciling the two credential planes (Vercel prod vs the local hermes fleet). Prevents the `APIFY_API_KEY` vs `APIFY_API_TOKEN` class of drift.
Create, verify, and promote Supabase schema changes via database branches. Every schema change/migration must be validated on a Supabase database branch before prod (see CLAUDE.md line 44).
Verify production Supabase schema before shipping code that touches it. Use BEFORE writing or reviewing ANY code that reads or writes a Supabase table in the Nexus — new queries, inserts, updates, CHECK-constrained values, RLS-dependent reads, cron data access — even one-line changes and even when the table "obviously" exists. Also use immediately when a query fails with "column does not exist", when inserts appear to succeed but produce no rows, or when generated TypeScript types disagree with runtime behaviour.
正在显示 SKILL.md
基于 SOC 职业分类
| id | delegation-planner |
| name | delegation-planner |
| type | Capability Uplift |
| version | 1.0.0 |
| created | 20/03/2026 |
| modified | 20/03/2026 |
| status | active |
| triggers | ["who should do this","which agent handles","delegate this","plan delegation","assign tasks","who handles","which specialist","route this task","assign work"] |
| description | > |
Purpose: Route every task to the correct agent layer before execution begins. Wrong routing wastes context budget and produces inferior output.
Use this skill when:
Route here when:
Do NOT route here:
Route here when:
Do NOT route here:
Route to Senior Engineering Agent when:
Route to Senior UI/UX Agent when:
Route to Senior QA / Production Agent when:
Route to Senior LMS Content Agent when:
Route to Senior Growth / Marketing Agent when:
Route here when:
Do NOT route here:
Ask: What kind of work is this?
If the task touches multiple domains (e.g., "add a feature with API + UI + tests"):
Produce the delegation plan table.
DELEGATION PLAN
═══════════════════════════════════════════════════
Task: [original task description]
Decomposed: [yes / no]
ROUTING
─────────────────
| # | Sub-task | Agent | Layer | Evidence Required | Gate |
|---|---------|-------|-------|------------------|------|
| 1 | [task] | [agent] | [1-4] | [what to return] | [gate condition] |
| 2 | [task] | [agent] | [1-4] | [what to return] | [gate condition — depends on #1] |
...
DEPENDENCY ORDER
─────────────────
[If tasks have dependencies: show order and gate conditions]
Step 1: Task #[N] must complete before Task #[N+1] begins
Gate: [exact condition that unlocks next step]
CONTEXT BUDGET
─────────────────
Orchestrator: [estimated tokens for coordination]
Each specialist: [estimated tokens per agent]
Total estimate: [rough total]
═══════════════════════════════════════════════════
Before finalising delegation plan:
| Failure | Recovery |
|---|---|
| Task is too vague to route | Ask: "What does success look like?" — activate outcome-translator |
| Task spans all 4 layers | Decompose into phases, route each phase separately |
| Wrong agent assigned (discovered mid-task) | Re-route immediately, don't continue in wrong layer |
| Sub-agent scope creeps to architectural decision | Pause sub-agent, escalate to orchestrator |
Task: "Add a protected dashboard page with user stats"
Delegation Plan:
| # | Sub-task | Agent | Layer | Evidence Required |
|---|---|---|---|---|
| 1 | API endpoint: GET /api/user/stats | Senior Engineering | 3A | curl output + type-check |
| 2 | Dashboard page component | Senior UI/UX | 3B | Screenshot + 0 TS errors |
| 3 | Auth middleware check | Senior Engineering | 3A | curl 401 without JWT |
| 4 | E2E test for dashboard | Senior QA | 3C | Playwright test output |
Dependency order: 1 → 3 → 2 → 4
Task: "Add dashboard" Response: "Sure, I'll add the dashboard." — No routing, no decomposition, no evidence plan.