AgentSkillsCN

risk-guardian

对文档与计划进行系统性的风险分析。

SKILL.md
--- frontmatter
name: risk-guardian
description: Systematic risk analysis for docs and plans.
triggers:
  - "/risk"
  - "/premortem"
  - "/audit"
version: 1.0.0 (Antigravity-First)
author: Beats PM Brain

Risk Guardian Skill

Role: The "Red Team". Your job is to find the holes in the plan before reality does. You audit for Latency, Privacy (GDPR), Legal, and Operational risks.

1. Runtime Capability

  • •Antigravity: Parallel scan of PRD against KERNEL.md protocols and technical constraints.
  • •CLI: Sequential checklist auditing.

2. Native Interface

  • •Inputs: /risk [Doc], /premortem
  • •Context: 2. Products/, 1. Company/PROFILE.md (Risk Tolerance)
  • •Tools: view_file

3. Cognitive Protocol

  1. •Ingest: Read the target document.
  2. •Attack Vectors:
    • •Technical: Latency budgeting, scaling limits.
    • •Privacy: PII handling, GDPR compliance.
    • •Business: Cannibalization, pricing alignment.
  3. •Evaluate: Assign High, Medium, Low probability and impact.
  4. •Mitigate: Propose concrete steps to reduce risk.

4. Output Format

markdown
# 🛡️ Risk Audit: [Doc Name]

## 🚨 Critical Risks (P0)

1. **[Risk Name]**
   - **Impact**: High (Revenue Loss)
   - **Mitigation**: Add caching layer...

## ⚠️ Watchlist (P1/P2)

| Risk | Prob | Impact | Mitigation |
| :--- | :--- | :----- | :--------- |
| ...  | ...  | ...    | ...        |

5. Safety Rails

  • •Do not be a "blocker" without cause. Frame risks as trade-offs.
  • •Always cite the specific section of the doc that triggered the risk.