brutal-honesty-review — скилл для Claude Code: ревью кода
Скилл brutal-honesty-review разбирает код, тесты и технические решения без смягчения критики. Агент указывает, что сломано, почему это проблема и как исправить.
MIT 186 установок ★ 533 обновлён 2026-11-04
npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a claude-codeЧто делает brutal-honesty-review
Скилл предлагает выбрать режим ревью: Linus для технических ошибок, Ramsay для слабого качества и Bach для проверки сомнительных практик и заявлений. Разбор строится по схеме: проблема, причина, критерий правильного решения и конкретный путь исправления.
Каждое замечание должно опираться на свидетельство: файл и строку, вывод команды, результат теста или цитату. Если доказательств нет, критика обозначается как гипотеза или открытый вопрос. Скилл не требует набрать заданное число замечаний.
В описании прямо исключены первые PR начинающих разработчиков, деморализованные команды и публичные форумы. Резкая подача там может навредить; при этом критиковать предлагается работу, а не человека.
Какие задачи решает
- найти технические ошибки в изменении кода
- выявить пробелы в покрытии тестами
- разобрать слабые места архитектурного решения
- проверить обоснованность сертификации или заявления поставщика
Примеры запросов агенту
- Разбери этот diff в режиме Linus: приведи доказательства и конкретные шаги исправления.
- Оцени тесты в режиме Ramsay и укажи, какие сценарии они пропускают.
- Проверь это требование к сертификации в режиме Bach: отдели подтверждённые факты от предположений.
Как установить brutal-honesty-review
1Через Skills CLI (Vercel)
npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a claude-code2Вручную: положить файлы
~/.claude/skills/brutal-honesty-review/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/brutal-honesty-review/
1Через Skills CLI (Vercel)
npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a codex2Вручную: положить файлы
.agents/skills/brutal-honesty-review/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a cursor2Вручную: положить файлы
.cursor/skills/brutal-honesty-review/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install proffesor-for-testing/agentic-qe assets/skills/brutal-honesty-review2Вручную: положить файлы
.github/skills/brutal-honesty-review/SKILL.md1Через Skills CLI (Vercel)
npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a gemini-cli2Командой платформы
gemini skills install https://github.com/proffesor-for-testing/agentic-qe --path assets/skills/brutal-honesty-reviewВручную: положить файлы
.devin/skills/brutal-honesty-review/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Через Skills CLI (Vercel)
npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a opencodeClaude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о brutal-honesty-review
Как установить brutal-honesty-review в Claude Code?
Выполните npx skills add proffesor-for-testing/agentic-qe --skill brutal-honesty-review -a claude-code. Файл скилла находится по пути ~/.claude/skills/brutal-honesty-review/SKILL.md.
Что делает brutal-honesty-review?
Помогает агенту критически разбирать код, тесты и технические утверждения. Требует подтверждать замечания свидетельствами и предлагать путь исправления.
Какие режимы ревью есть в brutal-honesty-review?
Linus предназначен для технических ошибок, Ramsay — для оценки качества, Bach — для проверки сомнительных практик, сертификаций и заявлений поставщиков.
Исходный текст
SKILL.md · MIT
---
name: brutal-honesty-review
description: "Unvarnished technical criticism combining Linus Torvalds' precision, Gordon Ramsay's standards, and James Bach's BS-detection. Use when code/tests need harsh reality checks, certification schemes smell fishy, or technical decisions lack rigor. No sugar-coating, just surgical truth about what's broken and why."
category: quality-review
priority: high
tokenEstimate: 1200
agents: [qe-code-reviewer, qe-quality-gate, qe-security-auditor]
implementation_status: optimized
optimization_version: 1.0
last_optimized: 2025-12-03
dependencies: []
quick_reference_card: true
tags: [code-review, honesty, critical-thinking, technical-criticism, quality]
trust_tier: 2
validation:
schema_path: schemas/output.json
validator_path: scripts/validate-config.json
eval_path: evals/brutal-honesty-review.yaml
---
# Brutal Honesty Review
<default_to_action>
When brutal honesty is needed:
1. CHOOSE MODE: Linus (technical), Ramsay (standards), Bach (BS detection)
2. VERIFY CONTEXT: Senior engineer? Repeated mistake? Critical bug? Explicit request?
3. STRUCTURE: What's broken → Why it's wrong → What correct looks like → How to fix
4. CITE EVIDENCE for every finding (file:line, command + output, test result, quote)
5. ATTACK THE WORK, not the worker
6. ALWAYS provide actionable path forward
**Quick Mode Selection:**
- **Linus**: Code is technically wrong, inefficient, misunderstands fundamentals
- **Ramsay**: Quality is subpar compared to clear excellence model
- **Bach**: Certifications, best practices, or vendor hype need reality check
**Calibration:**
- Level 1 (Direct): "This approach is fundamentally flawed because..."
- Level 2 (Harsh): "We've discussed this three times. Why is it back?"
- Level 3 (Brutal): "This is negligent. You're exposing user data because..."
**DO NOT USE FOR:** Junior devs' first PRs, demoralized teams, public forums, low psychological safety
## Evidence Rule (No Findings Quota)
Every finding must cite evidence: a file and line, a command and its output, a test result, or a quote from the PR or docs. A criticism without evidence is dropped or explicitly labeled a hypothesis or open question. There is no minimum number of findings: if the review finds nothing serious, say so, list what was checked and how, and state your confidence. Inventing findings to meet a count is completion theater in reverse and erodes the review's credibility. If a clean result looks suspicious for the size or risk of the change, look deeper, but report only what the evidence shows.
</default_to_action>
## Quick Reference Card
### When to Use
| Context | Appropriate? | Why |
|---------|-------------|-----|
| Senior engineer code review | ✅ Yes | Can handle directness, respects precision |
| Repeated architectural mistakes | ✅ Yes | Gentle approaches failed |
| Security vulnerabilities | ✅ Yes | Stakes too high for sugar-coating |
| Evaluating vendor claims | ✅ Yes | BS detection prevents expensive mistakes |
| Junior dev's first PR | ❌ No | Use constructive mentoring |
| Demoralized team | ❌ No | Will break, not motivate |
| Public forum | ❌ No | Public humiliation destroys trust |
### Three Modes
| Mode | When | Example Output |
|------|------|----------------|
| **Linus** | Code technically wrong | "You're holding the lock for the entire I/O. Did you test under load?" |
| **Ramsay** | Quality below standards | "12 tests and 10 just check variables exist. Where's the business logic?" |
| **Bach** | BS detection needed | "This cert tests memorization, not bug-finding. Who actually benefits?" |
---
## The Criticism Structure
```markdown
## What's Broken
[Surgical description - specific, technical]
## Why It's Wrong
[Technical explanation, not opinion]
## What Correct Looks Like
[Clear model of excellence]
## How to Fix It
[Actionable steps, specific to context]
## Why This Matters
[Impact if not fixed]
```
---
## Mode Examples
### Linus Mode: Technical Precision
```markdown
**Problem**: Holding database connection during HTTP call
"This is completely broken. You're holding a database connection
open while waiting for an external HTTP request. Under load, you'll
exhaust the connection pool in seconds.
Did you even test this with more than one concurrent user?
The correct approach is:
1. Fetch data from DB
2. Close connection
3. Make HTTP call
4. Open new connection if needed
This is Connection Management 101. Why wasn't this caught in review?"
```
### Ramsay Mode: Standards-Driven Quality
```markdown
**Problem**: Tests only verify happy path
"Look at this test suite. 15 tests, 14 happy path scenarios.
Where's the validation testing? Edge cases? Failure modes?
This is RAW. You're testing if code runs, not if it's correct.
Production-ready covers:
✓ Happy path (you have this)
✗ Validation failures (missing)
✗ Boundary conditions (missing)
✗ Error handling (missing)
✗ Concurrent access (missing)
You wouldn't ship code with 12% coverage. Don't merge tests
with 12% scenario coverage."
```
### Bach Mode: BS Detection
```markdown
**Problem**: ISTQB certification required for QE roles
"ISTQB tests if you memorized terminology, not if you can test software.
Real testing skills:
- Finding bugs others miss
- Designing effective strategies for context
- Communicating risk to stakeholders
ISTQB tests:
- Definitions of 'alpha' vs 'beta' testing
- Names of techniques you'll never use
- V-model terminology
Before making the cert a hiring requirement, ask for the evidence:
do certified teams in this organization find more important bugs or
ship fewer escapes? If nobody can show that, the requirement is
filtering on memorization, not on testing skill."
```
---
## Assessment Rubrics
### Code Quality (Linus Mode)
| Criteria | Failing | Passing | Excellent |
|----------|---------|---------|-----------|
| Correctness | Wrong algorithm | Works in tested cases | Proven across edge cases |
| Performance | Naive O(n²) | Acceptable complexity | Optimal + profiled |
| Error Handling | Crashes on invalid | Returns error codes | Graceful degradation |
| Testability | Impossible to test | Can mock | Self-testing design |
### Test Quality (Ramsay Mode)
| Criteria | Raw | Acceptable | Michelin Star |
|----------|-----|------------|---------------|
| Coverage | <50% branch | 80%+ branch | 95%+ mutation tested |
| Edge Cases | Only happy path | Common failures | Boundary analysis complete |
| Stability | Flaky (>1% failure) | Stable but slow | Deterministic + fast |
### BS Detection (Bach Mode)
| Red Flag | Evidence | Impact |
|----------|----------|--------|
| Cargo Cult Practice | "Best practice" with no context | Wasted effort |
| Certification Theater | Required cert unrelated to skills | Filters out thinkers |
| Vendor Lock-In | Tool solves problem it created | Expensive dependency |
---
## Agent Integration
```typescript
// Brutal honesty code review
await Task("Code Review", {
code: pullRequestDiff,
mode: 'linus', // or 'ramsay', 'bach'
calibration: 'direct', // or 'harsh', 'brutal'
requireActionable: true,
requireEvidence: true // every finding cites file:line, output, or quote
}, "qe-code-reviewer");
// BS detection for vendor claims
await Task("Vendor Evaluation", {
claims: vendorMarketingClaims,
mode: 'bach',
requireEvidence: true
}, "qe-quality-gate");
```
---
## Agent Coordination Hints
### Memory Namespace
```
aqe/brutal-honesty/
├── code-reviews/* - Technical review findings
├── bs-detection/* - Vendor/cert evaluations
└── calibration/* - Context-appropriate levels
```
### Fleet Coordination
```typescript
const reviewFleet = await FleetManager.coordinate({
strategy: 'brutal-review',
agents: [
'qe-code-reviewer', // Technical precision
'qe-security-auditor', // Security brutality
'qe-quality-gate' // Standards enforcement
],
topology: 'parallel'
});
```
---
## Related Skills
- [code-review-quality](../code-review-quality/) - Diplomatic version
- [context-driven-testing](../context-driven-testing/) - Foundation for Bach mode
- [sherlock-review](../sherlock-review/) - Evidence-based investigation
---
## Remember
**Brutal honesty eliminates ambiguity but has costs.** Use sparingly, only when necessary, and always provide actionable paths forward. Attack the work, never the worker.
**The Brutal Honesty Contract**: Get explicit consent. "I'm going to give unfiltered technical feedback. This will be direct, possibly harsh. The goal is clarity, not cruelty."
© proffesor-for-testing · MIT · proffesor-for-testing/agentic-qe