Files
assist/.claude/skills/kb-audit/scripts/kb_audit.py
zhaojie 0bee1c86fc feat: 新增 AI 指标报告助手与配置健康检查功能
- 新增 `ai-metrics-report` 技能,自动生成 AI 成功率、错误率与 Token 成本的综合报告,帮助评估智能助手表现。
- 新增 `config-audit` 技能,检查当前环境配置的完整性与可用性,输出健康检查报告,确保系统稳定运行。
- 相关脚本已实现,支持从项目根目录执行并输出结构化结果。
2026-02-11 00:59:55 +08:00

90 lines
2.8 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
知识库体检脚本
对 KnowledgeEntry 做简单统计,供 kb-audit Skill 调用。
"""
import sys
from datetime import datetime, timedelta
from pathlib import Path
def add_project_root_to_path():
# 假定脚本位于 .claude/skills/kb-audit/scripts/ 下
script_path = Path(__file__).resolve()
project_root = script_path.parents[4]
if str(project_root) not in sys.path:
sys.path.insert(0, str(project_root))
def main():
add_project_root_to_path()
from src.core.database import db_manager
from src.core.models import KnowledgeEntry
print("=== 知识库健康检查 ===\n")
with db_manager.get_session() as session:
total = session.query(KnowledgeEntry).count()
print(f"知识条目总数: {total}")
# 低置信度(<0.7
low_conf = (
session.query(KnowledgeEntry)
.filter(KnowledgeEntry.confidence_score.isnot(None))
.filter(KnowledgeEntry.confidence_score < 0.7)
.count()
)
print(f"低置信度条目数 (confidence_score < 0.7): {low_conf}")
# 使用次数极低usage_count < 3 或为 NULL
low_usage = (
session.query(KnowledgeEntry)
.filter(
(KnowledgeEntry.usage_count.is_(None))
| (KnowledgeEntry.usage_count < 3)
)
.count()
)
print(f"使用次数极低条目数 (usage_count < 3 或空): {low_usage}")
# 长期未更新(> 90 天)
cutoff = datetime.now() - timedelta(days=90)
old_entries = (
session.query(KnowledgeEntry)
.filter(
(KnowledgeEntry.updated_at.isnot(None))
& (KnowledgeEntry.updated_at < cutoff)
)
.count()
)
print(f"长期未更新条目数 (updated_at > 90 天未更新): {old_entries}")
print("\n示例问题条目(不含完整答案,仅展示前若干个):")
sample_entries = (
session.query(KnowledgeEntry)
.order_by(KnowledgeEntry.created_at.desc())
.limit(5)
.all()
)
for e in sample_entries:
q_preview = (e.question or "")[:40]
print(
f" ID={e.id}, category={e.category}, "
f"confidence={e.confidence_score}, usage={e.usage_count}, "
f"Q='{q_preview}...'"
)
print("\n提示:")
print(" - 建议优先审查低置信度且 usage_count 较高的条目;")
print(" - 对长期未更新且 usage_count 较高的条目,可考虑人工复查内容是否过时;")
print(" - 对 usage_count 极低且从未触发的条目,可考虑合并或归档。")
if __name__ == "__main__":
main()