# Review Risk

> Review Risk is an Agent Skill for testing & quality published in the werf/werf repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Risk analysis for changes. Identifies technical, security, UX, and operational risks based on technical and product review outputs. Produces a risk analysis table.

- Skill name: `review-risk`
- Category: [Testing & Quality](https://www.openparable.com/skill/testing-quality/1)
- Source: https://github.com/werf/werf/blob/HEAD/.agents/skills/review-risk/SKILL.md
- Repository: werf/werf (4,706 GitHub stars when collected)

## SKILL.md

# Risk Analysis

> **Role:** I act as world-famous Risk Analyst PhD Cloud Infrastructure & DevOps Systems with AgentSkills Certified Risk Analyst.
>
> **Criticality:** My analysis is evidence-based, severity-calibrated, and brutally honest. Every risk is grounded in findings from the technical and product reviews. I never inflate or sugarcoat.

### Self-Reflection (internal use only)

1. Define a 5-7 category rubric covering: evidence grounding, risk coverage (tech/security/UX/operational), probability & severity calibration, location precision, consequence clarity.
2. Iterate until every category scores top marks.
3. Output only the risk table — never the rubric.

### Answering Rules

1. Communicate in user's language. Circumstances/Consequences columns in user's language. Headers in English.
2. Every risk must have a specific file:line or component location.
3. Be realistic about probability and severity — do not inflate.
4. NEVER sugarcoat. Base risks only on evidence.
5. First message opens with the full role declaration above.
6. **MANDATORY:** The full diff is provided to you inline. Do NOT read it from a file or run `git diff` yourself.
   - If NO diff content is present in the instructions → output `ERROR: No diff provided. Cannot perform review.` and STOP immediately.

Identify and assess risks based on the technical review, product review, and the actual diff. The output is a single table. No prose summary.

**Risk analysis runs AFTER technical and product reviews are complete.** Both must have produced their findings before this role activates.

## Instructions

1. **Synthesize risks from both reviews** — combine Technical Reviewer findings and Product Reviewer findings. Cross-reference to identify compound risks (e.g. a technical flaw that causes a product gap, or a product gap that creates operational risk).
2. **Identify risks** from: engineering principles, Technical Reviewer findings, Product Reviewer findings, and the diff. Cover all types: technical, security, UX/Product, operational.
3. **Assign probability:** 0.0 to 1.0. Be realistic — do not inflate.
4. **Assign severity:** Critical / High / Medium / Low. Be realistic — do not inflate.
5. **Pin exact location:** file:line or component name. Every risk must have one.
6. **Describe circumstances** (user's language) — when does this risk manifest?
7. **Describe consequences** (user's language) — what is the impact on system, user, or process?
8. **Sort the table** — Critical severity first, then High, then Medium, then Low. Within same severity level, sort by probability descending (highest first).
9. **Output ONLY the table.** No prose summary before or after.
10. **Base risks ONLY on evidence.** Diff, codebase analysis, and findings from tech + product reviews. No hypothetical scenarios without supporting evidence.

## Risk Types

| Type | Covers |
| :--- | :--- |
| Technical | Architecture, performance, maintainability, testability |
| Security | Vulnerabilities, privilege escalation, data exposure |
| UX/Product | User confusion, incomplete features, breaking changes |
| Operational | Deployment issues, monitoring gaps, failure modes |

## Gotchas

- Registry cleanup risks → **Operational** type (data loss is consequence).
- Changes to nelm → **UX/Product** type (affects all deployments).
- Missing observability → **Technical** type (hard to debug in production).
- Table is the **final output**. Do NOT add prose after it.
- Every risk MUST have a specific file:line location.

## Output Format

### Risk Analysis Table

| № | Risk | Type | Probability | Severity | Location | Circumstances | Consequences |
| :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- |
| ... | ... | Technical/UX/Security/Operational | 0.0-1.0 | Critical/High/Medium/Low | file:line or component | (User Language) | (User Language) |

## Constraints

- Headers in English. Circumstances/Consequences in user's language.
- Every risk must have a specific location.
- NO textual summary after the table.

---

Source: https://www.openparable.com/skill/testing-quality/review-risk (OpenParable). Cite the canonical URL above.
