Security audit
Security checks for vulnerabilities and agentic risk
This skill openly aims to modify installed skills, but its write behavior is broader and less safely bounded than its own documentation claims.
Install only if you intentionally want a tool that can rewrite or augment other installed skills. Prefer using scan and propose first, run apply only on a small backed-up test skill directory, review diffs afterward, and avoid recording personal preferences unless you are comfortable keeping them in learned_patterns.json.
2. **提案（propose）**：把每个机会映射成安全补丁动作（仅追加/复制，从不删除/覆盖源），产出可审计的补丁计划。 3. **试应用（apply·沙箱）**：补丁先在临时副本上试跑，确认目标标记出现、文件不损坏后才允许落盘；任何异常都回滚。 4. **校验（verify）**：应用后重跑 scan，断言该机会已消除，否则判定补丁失败并保留原状。 5. **记忆（record）**：通过的补丁写入技能级 `self_improve_log.json`，并回写 evolver 的 `self_patches`，使整个生态的元改进轨迹可追溯。 ## 何时使用
changed = False
md = os.path.join(skill_dir, "SKILL.md")
txt = _read(md)
if issue == "no_self_evolve_section":
if "自进化学习系统" not in txt:
with open(md, "a", encoding="utf-8") as f:
f.write(SELF_EVOLVE_SECTION)No suspicious patterns detected.