一键导入
design-rules
Design phase rules for /designer command. Load at /designer startup (step 0). Covers spec quality, design checklist, approach evaluation criteria.
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。
菜单
Design phase rules for /designer command. Load at /designer startup (step 0). Covers spec quality, design checklist, approach evaluation criteria.
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
Task analysis and planning rules for /planner command. Load at /planner startup (step 0) or when /workflow enters Phase 1. Covers: task classification (7 types), S/M/L/XL complexity routing, code-researcher delegation (L/XL), plan documentation with full code examples.
Orchestration protocols for /workflow pipeline. Load at /workflow startup (step 0.1), then load individual protocols on-demand per event triggers. Covers: handoff contracts (4 phase-to-phase contracts + narrative casting), checkpoint format (session recovery), re-routing (complexity mismatch), pipeline metrics (completion tracking).
Project-local terse-output mode for /workflow pipeline. Cuts Messages token cost via concise prose. Loaded only at SessionStart by .claude/scripts/caveman-activate.sh — not auto-invoked on user prompts.
Review standards for code-reviewer agent. Loaded by explicit Read in the code-reviewer STARTUP step (disable-model-invocation blocks subagent preload). Covers: severity classification (BLOCKER/MAJOR/MINOR/NIT), decision matrix (APPROVED/APPROVED_WITH_COMMENTS/CHANGES_REQUESTED), auto-escalation rules, grep search patterns for automated checks.
Implementation rules and patterns for /coder command. Load at /coder startup (step 0) or when /workflow enters Phase 3. Covers: 5 CRITICAL rules (plan-only, import matrix, clean domain, no log+return, tests pass), evaluate protocol (PROCEED/REVISE/RETURN), dependency-ordered implementation.
Review standards for plan-reviewer agent. Loaded by explicit Read in the plan-reviewer STARTUP step (disable-model-invocation blocks subagent preload). Covers: architecture compliance (import matrix, domain purity), required plan sections validation, severity classification (BLOCKER/MAJOR/MINOR/NIT), decision matrix (APPROVED/NEEDS_CHANGES/REJECTED).
| name | design-rules |
| description | Design phase rules for /designer command. Load at /designer startup (step 0). Covers spec quality, design checklist, approach evaluation criteria. |
| disable-model-invocation | true |
Guidelines for the /designer command to produce high-quality design specs that reduce plan-review iterations.
Read this SKILL.md for overview. Supporting files loaded on-demand per phase.
| Criterion | Required | Check |
|---|---|---|
| Context describes current state | Yes | Not just "add X" but "currently Y exists, need X because Z" |
| Scope has IN and OUT | Yes | OUT items have explicit reasons |
| At least 2 approaches compared | Yes | With pros/cons for each |
| Selected approach has rationale | Yes | References constraints from requirements |
| Key decisions are numbered | Yes | Each has rationale and impact |
| Risks have severity and mitigation | Yes | HIGH risks must have concrete mitigation |
| Acceptance criteria are verifiable | Yes | Each can be checked as pass/fail |
| Anti-Pattern | Why Bad | Fix |
|---|---|---|
| "The obvious approach is..." | Skips exploration, may miss better options | Always compare at least 2 |
| Spec without OUT scope | Scope creep during planning | Explicitly list what's excluded |
| Vague acceptance criteria ("works well") | Can't verify | Make criteria concrete and testable |
| No risks identified | Every design has risks | Identify at least 1 per approach |
| Copying task description as context | No analysis | Describe current state, not just goal |
Cause: Task seems clear. Fix: Even "clear" tasks benefit from scope IN/OUT confirmation. At minimum, confirm scope.
Cause: Missing constraint not captured. Fix: Ask: "What constraint am I missing?" — don't generate more approaches without new information.