/deep-review

Comprehensive code review with full context analysis. Includes architecture review, test coverage, and impact analysis.

Phase 1: Gather Context (3 Parallel Haiku Agents)

Agent 1A - Change Summary:

1. Run `git status`, `git diff`, `git diff --staged`
2. If no changes → inform user and stop
3. Extract:
   - Files changed (list)
   - Languages detected (from extensions)
   - Line counts (additions/deletions)

Agent 1B - Project Context:

1. Find CLAUDE.md (root + directories with changes)
2. Read dependency files:
   - package.json / requirements.txt / go.mod / Cargo.toml
3. Identify project type and framework

Agent 1C - Related Files:

For each modified file, find:
- Files that import the modified file
- Files that the modified file imports
- Test files (foo.ts → foo.test.ts)

Phase 2: Load Agents (Progressive)

Only load agents relevant to detected context:

Condition	Load Agent
Always	@agents/bugs.md
Always	@agents/security.md
Always (deep)	@agents/architecture.md
CLAUDE.md exists	@agents/compliance.md
.ts/.tsx/.js/.jsx files	@agents/language-typescript.md
.py files	@agents/language-python.md

See @agents/index.md for full routing logic.

Phase 3: Deep Analysis (Parallel Sonnet Agents)

Launch loaded agents in parallel. Each agent:

1. Reads full file context + related files from Phase 1C
2. Analyzes only the diff (not pre-existing code)
3. Returns structured issues with diff-style fixes

Core Agents (always):

• @agents/bugs.md - Logic errors, null access, race conditions
• @agents/security.md - OWASP Top 10, injection, XSS, secrets
• @agents/architecture.md - Patterns, coupling, dependencies

Conditional Agents:

• @agents/compliance.md - If CLAUDE.md exists
• @agents/language-typescript.md - If TS/JS files
• @agents/language-python.md - If Python files

Additional Deep Analysis:

• Tests & Documentation Agent:
  • Do test files exist for modified code?
  • Do tests need updating for this change?
  • Are new public APIs missing tests?
  • Do README/docs need updates?

Phase 4: Impact Analysis (Sonnet Agent)

Using Phase 1C results, analyze:

• What other parts of codebase could be affected?
• Are there breaking changes to public APIs?
• Could this affect performance at scale?
• Are there database/schema implications?
• What's the blast radius if this has a bug?

Phase 5: Score & Filter (Haiku Agents)

For each issue, score confidence 0-100:

Factor	Points
In the diff (new code)	+20
Would cause failure	+30
In CLAUDE.md rules	+20
Senior engineer would flag	+20
Has ignore comment	-50

Apply filters from @templates/false-positive-rules.md:

• Filter issues with score < 70 (lower threshold for deep review)
• Track filtered count by reason

Phase 6: Present Results

Format output using @templates/output-format.md:

## Deep Code Review

**Summary:** Reviewed X files, Y lines changed

| Found | Reported | Filtered |
|-------|----------|----------|
| total | ≥70 score | <70 score |

### Critical (95-100) 🔴
[Issues with diff-style fixes]

### Warning (80-94) 🟠  
[Issues with diff-style fixes]

### Medium (70-79) 🟡
[Issues with diff-style fixes]

### Filtered Issues 🔇
[Count by reason, expandable details]

### Architectural Notes 📐
- Pattern consistency: ✅/⚠️/❌
- Test coverage: ✅/⚠️/❌
- Documentation: ✅/⚠️/❌
- Dependencies: ✅/⚠️/❌

### Impact Analysis 💥
- Affected files: [list]
- Blast radius: [scope]
- Breaking changes: [yes/no]

Differences from Quick Review

Aspect	Quick Review	Deep Review
Threshold	≥80	≥70
Architecture	❌	✅
Impact Analysis	❌	✅
Test Coverage	❌	✅
Related Files	❌	✅
Medium Priority	❌	✅