← 返回
安全合规 中文

Memory Audit Guardian

Weekly memory governance audit for OpenClaw. Use when user asks to audit/optimize memory quality, reduce token overhead, verify MEMORY/TOOLS/AGENTS role boun...
每周对 OpenClaw 进行记忆治理审计。当用户要求审计/优化记忆质量、降低 token 开销、验证 MEMORY/TOOLS/AGENTS 角色边界时使用。
markisadesigner1992-eng
安全合规 clawhub v1.0.0 1 版本 100000 Key: 无需
★ 1
Stars
📥 631
下载
💾 11
安装
1
版本
#audit#automation#governance#latest#memory#qmd

概述

Memory Audit Guardian

Run a structured memory-system audit and output an actionable weekly report.

Audit Scope

  1. File-role boundaries
    • SOUL: persona only
    • USER: user profile only
    • MEMORY: durable high-value facts only
    • daily memory: event logs
    • TOOLS: execution rules
    • AGENTS: governance/policy
  2. Size & token budget
    • Core files total target: 8-10KB (soft target)
    • MEMORY.md warning threshold: >3KB
  3. QMD routing quality
    • Ensure MEMORY contains only QMD entry + protocol + 20-40 anchors
    • Reject full keyword dumps in MEMORY
  4. Retrieval discipline
    • top-k 3-5 snippets
    • no full-document dump unless explicitly requested

Procedure

  1. Read core files: SOUL.md, USER.md, MEMORY.md, TOOLS.md, AGENTS.md
  2. Compute file sizes and detect threshold breaches
  3. Detect duplication/conflict across files
  4. Check MEMORY for noise patterns (long lists, stale one-offs)
  5. Verify QMD routing fields exist and are minimal
  6. Produce report with:
    • score (A/B/C)
    • risks
    • fixes this week
    • add-3/remove-3 keyword suggestions

Output Format

  • Executive summary (5 lines max)
  • Findings:
  • role-boundary issues
  • size/token risks
  • QMD routing issues
  • Action plan (this week)
  • Report path suggestion: memory/memory-audit-YYYY-WW.md

Guardrails

  • Do not overwrite user files without explicit request.
  • Prefer append/update with minimal edits.
  • Keep MEMORY concise; move detail to references or daily logs.

版本历史

共 1 个版本

  • v1.0.0 当前
    2026-03-30 05:49 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

ai-intelligence

Memory Sync Protocol

markisadesigner1992-eng
通过统一工作流,在 TOOLS.md、MEMORY.md、AGENTS.md 及 memory/YYYY-MM-DD.md 之间同步持久化行为或偏好变更。
★ 2 📥 869
security-compliance

OpenClaw Backup

alex3alex
备份与恢复 OpenClaw 数据。适用于创建备份、设置自动备份计划、从备份恢复或管理备份轮转。处理 ~/.openclaw 目录归档并包含适当的排除规则。
★ 89 📥 30,586
security-compliance

Skill Vetter

spclaudehome
AI智能体技能安全预审工具。安装ClawdHub、GitHub等来源技能前,检查风险信号、权限范围及可疑模式。
★ 1,210 📥 266,150