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 skills add Stanshy/AgentHub --skill harness-auditgit clone --depth 1 https://github.com/Stanshy/AgentHubWrote 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/stanshy/agenthub/harness-audit)<a href="https://agentmods.dev/skills/stanshy/agenthub/harness-audit"><img src="https://agentmods.dev/badge/skills/stanshy/agenthub/harness-audit.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 | $0.00015 | $0.00826 |
| Opus 5 | $0.00008 | $0.00413 |
| Sonnet 5 | $0.00003 | $0.00165 |
| Haiku 4.5 | $0.00002 | $0.00083 |
Grade A, and why
harness-audit 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 8d 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.
What it actually says
Harness 健康審計
週期性檢查專案 Harness（CLAUDE.md + Hook + Skill + Knowledge）的健康狀態。
使用方式
/harness-audit
參數
無（操作當前專案）
檢查項目（七大原則）
原則 1: 上下文架構（給地圖不給百科全書）
- CLAUDE.md 行數 ≤ 100（超過 = 警告）
- CLAUDE.md 有文件索引且索引完整
- .knowledge/ 文件都在索引中登記
原則 2: 架構約束（用工具強制不靠 prompt）
- 掃描 CLAUDE.md 中的「禁止」「不得」等文字規則
- 檢查是否有對應 Hook 強制（.claude/settings.json）
- 未被 Hook 覆蓋的文字規則列出建議
原則 3: 知識層級（公司 vs 專案）
- .knowledge/company-rules.md 存在
- .knowledge/team-workflow.md 存在
- 專案級規範與公司級不衝突
原則 4: 技能即流程（Skill 覆蓋）
- ~/.claude/commands/ 或 .claude/commands/ 目錄存在且有 skill 檔案
- 掃描已部署的 Skill 數量
- 列出未部署但可用的公司 Skill（對照 AgentHub skill-templates/）
原則 5: Hook 健康
- .claude/settings.json 存在且 JSON 合法
- Stop hook 存在
- PreToolUse hook 存在（forbidden-commands）
- Hook 腳本檔案存在且可執行
原則 6: 文件新鮮度
- .knowledge/ 下各文件最後修改日期
- 超過 30 天未更新的文件標記為「可能過時」
原則 7: 整體評分
- 每個原則 0-2 分（0=缺失, 1=部分, 2=完善）
- 總分 /14，≥10 = 健康，7-9 = 需改善，<7 = 警告
輸出格式
# Harness 健康審計報告
**專案**: {project-name}
**日期**: {today}
**總分**: {score}/14
## 評分明細
| 原則 | 分數 | 說明 |
|------|------|------|
| 1. 上下文架構 | {0-2} | {detail} |
| 2. 架構約束 | {0-2} | {detail} |
| 3. 知識層級 | {0-2} | {detail} |
| 4. 技能覆蓋 | {0-2} | {detail} |
| 5. Hook 健康 | {0-2} | {detail} |
| 6. 文件新鮮度 | {0-2} | {detail} |
| 7. 整體完整性 | {0-2} | {detail} |
## 行動項目
1. {highest priority action}
2. {second priority action}
...
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.
- 8d ago First seen · 80 lines · 15 tokens per session scan A 994d2a716121
harness-audit is a skill published in the GitHub repository Stanshy/AgentHub (201 stars, last pushed 5mo ago), licensed MIT. It adds 15 tokens to every session and 826 once invoked, about $0.0001 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
cross-campaign
Discover and reference other camps, projects, and files across camp boundaries. Use when the user mentions another camp or campaign by name, references work done "in another project/camp", or needs to find/copy/compare code across camps.
issue-triage
3-phase issue backlog management with audit, deep analysis, and validated triage actions. Use when triaging GitHub issues, sorting bug reports, cleaning up stale tickets, or detecting duplicate issues. Args: 'all' to analyze all, issue numbers to focus (e.g. '42 57'), 'en'/'fr' for language, no arg = audit only.
review-pr
Perform a comprehensive code review of a pull request.
guide-recap
Transform CHANGELOG entries into social content (LinkedIn, Twitter/X, Newsletter, Slack) in FR + EN. Use after releases or weekly to generate release notes, announcements, social media posts, or recap summaries from guide updates.
release-notes-generator
Generate release notes in 3 formats (CHANGELOG.md, PR body, Slack announcement) from git commits. Automatically categorizes changes and converts technical language to user-friendly messaging. Use for releases, changelogs, version notes, what's new summaries, or ship announcements.
talk-pipeline
Orchestrates the complete talk preparation pipeline from raw material to revision sheets, running 6 stages in sequence with human-in-the-loop checkpoints for REX or Concept mode talks. Use when starting a new talk pipeline, resuming a pipeline from a specific stage, or running the full end-to-end preparation workflow.