📊 统一分析日报

📅 2026-08-06 | Blueprint(30条日志) + Hermes(17个会话) | 自动生成 00:32

🔴 5.1%
CC委派率
17
会话数
30
Blueprint日志
36.8%
错误率
$0.00
成本
1
交叉发现

📊 全部发现 (17项) — 按域分类

严重度发现类型
BPhigh僵尸改进项: patch工具替换HTML时意外吞掉了相邻标签(
zombie_improvement
BPhigh僵尸改进项: patch工具替换HTML时意外吞掉相邻的
zombie_improvement
BPhigh僵尸改进项: Playwright 测试大面积失败但错误信息误导:8099 端口被 wechat-qa-bot 项目的 4 个残留 hzombie_improvement
BPhigh僵尸改进项: python http.server 服务了错误文件:项目根目录残留旧 index.html(835KB),而新构建的 zombie_improvement
BPhigh僵尸改进项: Playwright 测试 locator('h1') strict mode violation:页面有 2 个 h1zombie_improvement
BPhigh僵尸改进项: Claude Code 视觉检查 OOM(exit -9 × 2):一次性让 claude -p 读 5 张大截图(13zombie_improvement
BPhigh僵尸改进项: 分析器成功率口径错误:未标注overall.success的日志被算作失败,导致成功率显示45%而非实际90%zombie_improvement
BPhighPhase3_implementation 问题率 7/11 (64%)phase_bottleneck
BPhighPhase4_retrospective 问题率 2/2 (100%)phase_bottleneck
BPhighInvestigate 'patch' (2 failures). Consider adding retry logic, fallback, or replsuggestion
BPhighPhase3_implementation has 64% issue rate. Review phase procedures — possible missuggestion
BPhighPhase4_retrospective has 100% issue rate. Review phase procedures — possible missuggestion
BPmediumgate3 failed 1 times. Analyze failure patterns — might need pre-gate checklist osuggestion
BPmediumTop user correction cause: 'design_flaw'. Review related skills for outdated inssuggestion
HEhighCC 委派率 5.1% (目标>50%)cc_delegation_rate

🔧 Hermes 运营分析 数据源: state.db

编码方式次数占比
Hermes 直接编码24094.9%
Claude Code135.1%

⚠️ CC违规 (1)

06:00(无标题)6次

❌ 错误分布 (276个)

terminal206
execute_code16
read_file14
memory12
browser_navigate12

📋 Blueprint 流程分析 数据源: JSONL日志

30总执行数
10/11 (91%)成功率
7僵尸改进项
0待解决问题

🔴 僵尸改进项 (7)

⚠️ patch工具替换HTML时意外吞掉了相邻标签(
skill_updated=false
⚠️ patch工具替换HTML时意外吞掉相邻的
skill_updated=false
⚠️ Playwright 测试大面积失败但错误信息误导:8099 端口被 wechat-qa-bot 项目的 4 个残留 h skill_updated=false
⚠️ python http.server 服务了错误文件:项目根目录残留旧 index.html(835KB),而新构建的 skill_updated=false
⚠️ Playwright 测试 locator('h1') strict mode violation:页面有 2 个 h1 skill_updated=false

💡 改进建议

high Investigate 'patch' (2 failures). Consider adding retry logic, fallback, or replacing with alternati
high Phase3_implementation has 64% issue rate. Review phase procedures — possible missing checks or uncle
high Phase4_retrospective has 100% issue rate. Review phase procedures — possible missing checks or uncle
medium gate3 failed 1 times. Analyze failure patterns — might need pre-gate checklist or automated pre-chec
medium Top user correction cause: 'design_flaw'. Review related skills for outdated instructions.

🔗 交叉发现 需双轨修复

以下发现同时影响 Blueprint 和 Hermes,需在两处分别改进

CC 超时=0 OOM=2 (Blueprint日志)
Blueprint 影响 Blueprint prompt 策略 + Hermes CC 调用参数配置
Hermes 影响 Hermes 工具调用 + Blueprint prompt 策略