mirror of
https://github.com/catlog22/Claude-Code-Workflow.git
synced 2026-02-14 02:42:04 +08:00
从子目录执行 skill 时,相对路径 .workflow/ 会导致产物落到错误位置。
通过 git rev-parse --show-toplevel || pwd 检测项目根目录,
所有 .workflow/ 路径引用统一加上 {projectRoot} 前缀确保路径正确。
涉及 72 个文件,覆盖 20+ 个 skill。
4.8 KiB
4.8 KiB
Phase 5: Review Completion
Source: Shared from
commands/workflow/review-session-cycle.md+commands/workflow/review-module-cycle.mdPhase 5
Overview
Finalize review state, generate completion statistics, and optionally prompt for automated fix pipeline.
Execution Steps
Step 5.1: Finalize State
Phase 5 Orchestrator Responsibilities:
- Finalize review-progress.json with completion statistics
- Update review-state.json with completion_time and phase=complete
- Progress tracking: Mark all tasks done
review-state.json updates:
{
"phase": "complete",
"completion_time": "2025-01-25T15:00:00Z",
"next_action": "none"
}
review-progress.json updates:
{
"phase": "complete",
"overall_percent": 100,
"completion_time": "2025-01-25T15:00:00Z",
"final_severity_distribution": {
"critical": 0,
"high": 3,
"medium": 12,
"low": 8
}
}
Step 5.2: Evaluate Completion Status
Full Success:
- All dimensions reviewed
- Critical findings = 0
- High findings <= 5
- Action: Generate final report, mark phase=complete
Partial Success:
- All dimensions reviewed
- Max iterations reached
- Still have critical/high findings
- Action: Generate report with warnings, recommend follow-up
Step 5.3: Progress Tracking Completion
Update progress tracking to reflect all phases completed:
Phase 1: Discovery & Initialization → completed
Phase 2: Parallel Reviews (7 dimensions) → completed
→ Security review → completed
→ Architecture review → completed
→ Quality review → completed
... other dimensions as sub-items
Phase 3: Aggregation → completed
Phase 4: Deep-dive → completed
Phase 5: Completion → completed
Step 5.4: Fix Pipeline Prompt
- Ask user: "Run automated fixes on findings? [Y/n]"
- If confirmed AND --fix flag: Continue to Phase 6
- Display summary of findings by severity:
Review Complete - Summary:
Critical: 0 High: 3 Medium: 12 Low: 8
Total findings: 23
Dimensions reviewed: 7/7
Iterations completed: 2/3
Run automated fixes on findings? [Y/n]
Completion Conditions
Full Success:
- All dimensions reviewed
- Critical findings = 0
- High findings <= 5
- Action: Generate final report, mark phase=complete
Partial Success:
- All dimensions reviewed
- Max iterations reached
- Still have critical/high findings
- Action: Generate report with warnings, recommend follow-up
Error Handling Reference
Phase-Level Error Matrix
| Phase | Error | Blocking? | Action |
|---|---|---|---|
| Phase 1 | Invalid path pattern / Session not found | Yes | Error and exit |
| Phase 1 | No files matched / No completed tasks | Yes | Error and exit |
| Phase 1 | Files not readable / No changed files | Yes | Error and exit |
| Phase 2 | Single dimension fails | No | Log warning, continue other dimensions |
| Phase 2 | All dimensions fail | Yes | Error and exit |
| Phase 3 | Missing dimension JSON | No | Skip in aggregation, log warning |
| Phase 4 | Deep-dive agent fails | No | Skip finding, continue others |
| Phase 4 | Max iterations reached | No | Generate partial report |
CLI Fallback Chain
Gemini -> Qwen -> Codex -> degraded mode
Fallback Triggers
- HTTP 429, 5xx errors, connection timeout
- Invalid JSON output (parse error, missing required fields)
- Low confidence score < 0.4
- Analysis too brief (< 100 words in report)
Fallback Behavior
- On trigger: Retry with next tool in chain
- After Codex fails: Enter degraded mode (skip analysis, log error)
- Degraded mode: Continue workflow with available results
Best Practices
- Start Specific: Begin with focused module patterns for faster results
- Expand Gradually: Add more modules based on initial findings
- Use Glob Wisely:
src/auth/**is more efficient thansrc/**with lots of irrelevant files - Trust Aggregation Logic: Auto-selection based on proven heuristics
- Monitor Logs: Check reports/ directory for CLI analysis insights
Related Commands
View Review Progress
Use ccw view to open the review dashboard in browser:
ccw view
Automated Fix Workflow
After completing a review, use the generated findings JSON for automated fixing:
# Step 1: Complete review (this skill)
review-cycle src/auth/**
# OR
review-cycle
# Step 2: Run automated fixes using dimension findings
review-cycle --fix ${projectRoot}/.workflow/active/WFS-{session-id}/.review/
Output
- State: review-state.json (phase=complete), review-progress.json (final)
- Decision: fix pipeline or end
Next Phase
- If fix requested: Phase 6: Fix Discovery & Batching
- Else: Workflow complete