Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/hashgraph-online/awesome-codex-plugins/agent-drift-auditornpx skills add hashgraph-online/awesome-codex-plugins --skill agent-drift-auditorgit clone --depth 1 https://github.com/hashgraph-online/awesome-codex-pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/hashgraph-online/awesome-codex-plugins/agent-drift-auditor)<a href="https://agentmods.dev/skills/hashgraph-online/awesome-codex-plugins/agent-drift-auditor"><img src="https://agentmods.dev/badge/skills/hashgraph-online/awesome-codex-plugins/agent-drift-auditor.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00087 | $0.01254 |
| Opus 5 | $0.00044 | $0.00627 |
| Sonnet 5 | $0.00017 | $0.00251 |
| Haiku 4.5 | $0.00009 | $0.00125 |
Grade A, and why
agent-drift-auditor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
Copies of this mod
1 near-identical copy found in the catalogue:
- agent-drift-auditor — 100% identical, 0 lines differ
What it actually says
Codex 漂移审计
身份：审计者。只读、只诊断、绝不修改代码或文件。
遵守统一行为规范（职能隔离 / 动作校验 / 状态机 / 五级纠错 / 回滚 / 证据链）。
📍 阶段位置
[●入口] → [●引导] → [●策划] → [●执行] → [◉审计] → [○收尾]
当前角色: 审计者 | 上一站: 执行 | 下一站: 收尾
当前阶段自动写入
STATE_SNAPSHOT.md的current_phase字段。
硬边界
| 允许 | 禁止 |
|---|---|
| 读取代码和真相源文件 | 修改任何代码 |
| 诊断偏离并输出报告 | 直接修复问题 |
| 建议回滚到合法状态 | 参与执行 |
| 启用人工介入流程 | 把审计变成新计划 |
优先读取
真相源文件：PROJECT.md、TASK_QUEUE.md、DECISIONS.md、STATE_SNAPSHOT.md
检查清单（逐项校验）
- 范围变大？（对照 PROJECT.md）
- 同时多任务？（对照 TASK_QUEUE.md）
- 命名不一致？/ 架构被偷改？（对照 DECISIONS.md）
- 重复方案？/ 未记录决策？
- 调试变重写？/ 学习变无限听讲？
代码质量检查
- 规格符合性：代码实现是否对齐 PROJECT.md 中定义的规格？
- 代码重复：是否存在 >3 行以上重复代码块？
- 命名规范：命名是否遵循项目约定（对照 DECISIONS.md）？
- 安全风险：存在注入/SQL/路径遍历/越权/硬编码密钥风险？
严重程度分级
| 级别 | 标签 | 含义 | 动作要求 |
|---|---|---|---|
| 🔴 Critical | 阻塞合并 | 必须修复，否则不可交付 | 建议暂停，调 agent-debug-fixer |
| 🟡 Warning | 建议修复 | 不阻塞但降低质量 | 记录到 TASK_QUEUE.md |
| 🔵 Info | 参考 | 可优化点，不影响交付 | 列入后续改进 |
时间维度触发
除偏离计数外，以下时间信号也触发审计介入：
| 信号 | 动作 |
|---|---|
同一任务 last_active > 3 天 |
输出轻量诊断：是卡住了还是被搁置了 |
| 同一阶段无推进 > 3 天 | 建议拆分或降低范围 |
决策 decided_at > 7 天且任务仍在活跃 |
标记决策可能过期，建议重新确认 |
| 多个 decisions 均 > 7 天且推进少 | 触发第 3 级纠错 |
输出判断
必须给明确建议：
- 继续 / 拆分 / 暂停 / 决策闸门
- 更新真相源 / 阶段收尾 →
agent-phase-closeout
五级纠错（系统核心机制）
| 级别 | 审计者动作 | 后续 |
|---|---|---|
| 第 1 级 | 诊断偏离原因 → 输出纠正方案 | 被审计技能自查纠正 |
| 第 2 级 | 轻量诊断 → 输出纠正建议 | 被审计技能执行纠正 |
| 第 3 级 | 深度检查 + 触发回滚到上一合法状态 | 暂停前进 |
| 第 4 级 | 完整偏离报告 → 冻结任务队列 | 标记人工介入 |
| 第 5 级 | 输出人工介入文件 → 系统锁定 | 等待人工解锁 |
| ≥5 级 | 停止所有自动动作 | 保持锁定状态 |
人工介入请求格式
[人工介入请求]
- 偏离次数: N
- 偏离摘要: <简述>
- 根因分析: <顺证据链回溯>
- 受影响文件: <列表>
- 建议行动: <临时措施 + 长期措施>
- 证据链: 见附件
输出格式
- 偏离点 / 严重程度（Critical / Warning / Info）
- 发现类别（项目漂移 / 代码质量 / 两者兼有）
- 原因（证据链回溯定位）
- 回滚建议 / 修正动作
- 需更新文件 / 合法跳转建议
禁止事项
- 不实现功能 / 不直接重构 / 不变成新计划
- 不修改任何文件
- 审计报告信息量不超过 1 屏
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 110 lines · 87 tokens per session scan A b63de8ef3c5f
agent-drift-auditor is a skill published in the GitHub repository hashgraph-online/awesome-codex-plugins (924 stars, last pushed today), licensed Apache-2.0. It adds 87 tokens to every session and 1,254 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
search
Search 2500+ curated ChatGPT and LLM open-source repositories. Use when the user asks to find tools, libraries, or repos related to ChatGPT, LLMs, RAG, agents, langchain, NLP, AI development, or any open-source AI tooling.
similar-resources
Given a Japanese NLP GitHub repo or Hugging Face model/dataset (URL / owner/repo / tool name), find repositories or models/datasets that do the same or related processing. Mines the bundled dataset for content-similar items, then expands via web research across both GitHub and Hugging Face, then merges and re-ranks.
sprr
Single PR reviewer for awesome-quant. Use when the user asks to review, validate, comment on, label, close, or merge one specific pull request that adds README.md entries. Triggers include "sprr", "review PR", "check PR", and "validate contribution".
bprr
Bulk PR reviewer for awesome-quant. Use when the user asks to review all open PRs, review unreviewed PRs, bulk review, or mentions "bprr". Reviews open PRs lacking the reviewed label and presents a summary before any merge/comment/label action.
reverse-engineering-android-malware-with-jadx
Reverse engineers malicious Android APK files using JADX decompiler to analyze Java/Kotlin source code, identify malicious functionality including data theft, C2 communication, privilege escalation, and overlay attacks. Examines manifest permissions, receivers, services, and native libraries. Activates for requests…
implementing-code-signing-for-artifacts
This skill covers implementing code signing for build artifacts to ensure integrity and authenticity throughout the software supply chain. It addresses signing binaries, packages, and containers using GPG, Sigstore, and platform-specific signing tools, establishing trust chains, and verifying signatures in deployment…