Index of every page on this site: llms.txt
All skills
cat-xierluo avatar

/self-evolve

@5fb69a0
Run it yourself

扫描本机 Claude Code、Codex、WorkBuddy、CodeBuddy、ZCode 等 Agent 会话,保留初始化和增量处理记录,低成本初筛后由当前 Agent 提炼经验、迭代与 Skill 沉淀建议,生成可独立打开的 HTML 报告。也用于复盘当前对话、接续每周沉淀或检查扫描覆盖。无需单独配置模型 API Key;不自动应用建议或安装 Skill。

Use this Skill: https://skilld.dev/gh/cat-xierluo/legal-skills/self-evolve

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

references17-sediment-reports.md

≈790 tokens on demand. Your agent reads this file only when SKILL.md points to it.

沉淀报告(panel「沉淀报告」tab)

定位

review/inbox/ 下 self-evolve 写出的脱敏 md 报告(迭代周报 / deep-analysis 等),以阅读视图在 panel 第 4 个 tab「沉淀报告」统一展示。不再「手动翻 md」。

参照 info-evaluator 的运营模式(assets/preferences/README.md:「工作台和脚本都必须通过 scripts/preference_ledger.py 访问,不直接编辑文件」),self-evolve 的沉淀报告也走单一收集入口:render_panel.collect_review_documents()。

来源

产物 来源脚本 频率
iteration/YYMMDD*_weekly-iteration.md scripts/iteration_report.py weekly
deep-analysis/YYMMDD*_deep-analysis.md scripts/deep_dream.py --execute weekly(可选,默认 enabled: false)
YYMMDD*_skill_upgrade_*.md scripts/proposal_store 历史 legacy 散落
YYMMDD*_article_*.md scripts/proposal_store 历史 legacy 散落

新报告统一由各自写脚本经 privacy_gate.write_review_artifact 落盘,不在 panel 直接编辑。

安全

  • 所有 md 由 write_review_artifact 写入,已经过 privacy_gate.sanitize_text + validate_untrusted_source(原始路径/邮箱/手机/身份证/银行卡/组织名/prompt 注入/secret 全替换/拒绝)。
  • panel 前端再过 render_panel.safe_text(validate_untrusted_source + validate_text)第二层防护。
  • 单文件 ≤ 20000 字(safe_text max_chars),<details> 展开 + 滚动,避免一次性渲染大文档。
  • 不直接编辑 review/inbox/ 下的 md——通过写脚本追加。

与其他审阅的关系

Tab 对象 决定语义
Review Inbox proposals(json) 接受/拒绝/部分接受/观察
偏好审阅 workmode 偏好(json) 接受/观察/拒绝
沉淀报告 md 报告 阅读视图(MVP 不带决定)
Cron 任务 scheduled_tasks 读

沉淀报告是 digest,不是决定项(无 accept/reject 语义)。如需校准「提取是否精准」,是下一版(1.7.x)校准维度。

访问

  • panel 启动后顶部第 4 个 tab「沉淀报告」。
  • API:GET /api/review-documents[?kind=iteration&sort=newest] → {documents:[...], total, metrics:{total, this_week, by_kind}}(独立懒加载,不进 build_panel_data,避免大目录扫描拖累主面板)。
  • 收集入口:scripts/render_panel.collect_review_documents(kind=None, sort="newest", limit=50)。
  • 单一权威入口原则(参 info-evaluator README):其他脚本/前端不应直接 rglob review/inbox/,统一经 collect_review_documents()。

Metrics

panel 顶部展示:

  • total:全集计数(不被 kind/sort/limit 影响)
  • this_week:近 7 天 mtime 的报告数
  • by_kind:{iteration: N, deep-analysis: M, skill_upgrade: K, article: ..., document: ...}

未来(1.7.x)

  • 校准维度:每条归因 sample 可标记「准确 / 不准确 / 不确定」,反馈进 runtime/sediment-calibration/calibration.jsonl(append-only),panel 加校准交互。验证 self-evolve 提取质量有反馈闭环。
  • 报告摘要结构化:iteration_report 产出 JSON 摘要 + md 双产物,panel 可跨报告聚合。

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning18d3 checks · Risk SAFE
  • Gen Agent Trust Hub18d

    The skill is a tool for analyzing local AI agent session history to suggest improvements. It naturally requires access to local logs and performs local script execution, but it includes robust privacy gating and instructions to mitigate risks from untrusted data.

  • Socket18d

    No alerts

  • Snyk18d

    Risk: MEDIUM · 1 issue

Signed by skilld at 5fb69a0. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 4 weeks ago
Other metadata
metadata
{
  "version": "1.28.1",
  "homepage": "https://github.com/cat-xierluo/legal-skills",
  "author": "杨卫薪律师(微信ywxlaw)"
}

README badge

README badge for cat-xierluo/legal-skills/self-evolve