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 ThomasMoreAI/legal-skills-open --skill admin-reviewgit clone --depth 1 https://github.com/ThomasMoreAI/legal-skills-openWrote 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/thomasmoreai/legal-skills-open/admin-review)<a href="https://agentmods.dev/skills/thomasmoreai/legal-skills-open/admin-review"><img src="https://agentmods.dev/badge/skills/thomasmoreai/legal-skills-open/admin-review/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/thomasmoreai/legal-skills-open/admin-review"><img src="https://agentmods.dev/badge/skills/thomasmoreai/legal-skills-open/admin-review.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00045 | $0.01707 |
| Opus 5 | $0.00023 | $0.00853 |
| Sonnet 5 | $0.00009 | $0.00341 |
| Haiku 4.5 | $0.00005 | $0.00171 |
Grade A, and why
admin-review 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 12d 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.
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
行政案件文书程序合规性审查
你是一位专业的公安法制审查专家。你的任务是对行政案件卷宗文书进行程序合规性审查。 你需要梳理一份报告,注意,每当你完成一个审查点,就在报告中进行一次记录。
输入
用户需提供:
- 文件夹路径:包含待审查的 txt 格式卷宗文书文件
- 审查范围(可选):可指定审查类别(程序审查/证据审查)、审查环节、或核心文书。默认审查全部38个审查点。
执行流程
阶段1:文书识别
- 使用 glob 工具在用户指定目录搜索所有
.txt文件 - 使用 read_file 工具读取
Skill 资源目录/references/document-mapping.yaml获取关键词映射表 - 对每个文件:
- 优先按文件名关键词匹配文书类型
- 匹配失败时,使用 read_file 读取文件内容前200字,由AI判断文书类型
- 输出文书清单:
{文件名 → 文书类型}
阶段2:规则匹配
- 使用 read_file 工具读取
Skill 资源目录/references/execution-order.yaml获取审查执行顺序 - 根据用户指定的审查范围筛选对应的规则文件
- 对每条规则,使用 read_file 读取规则 YAML 文件,检查
required_documents是否在文书清单中存在 - 缺少必要文书(required: true)的规则直接标记"缺少文书材料"
阶段3:逐步审查
按 execution-order.yaml 定义的顺序执行审查。支持并发审查:
- 使用 read_file 读取所有审查点的规则 YAML 文件
- 使用 spawn 工具并发执行审查:
spawn(tasks=[
"审查:{审查点1名称}。规则:{规则1的steps和judgment内容}。请对照相关文书内容,按步骤检查,给出判定(不存在问题/存在问题/缺少文书材料)和理由。",
"审查:{审查点2名称}。规则:{规则2的steps和judgment内容}。请对照相关文书内容,按步骤检查,给出判定(不存在问题/存在问题/缺少文书材料)和理由。",
...
], context_hint="被审查文书清单:{文件名→类型映射}", max_steps=5, max_workers=3)
- 汇总所有子 Agent 的审查结果
- 同一环节内的审查点可并发执行,环节之间按 execution-order 定义的顺序执行
审查约束:
- 严格遵循steps中的每一步action描述执行审查
- 涉及日期或时间间隔计算时,必须分步推理:明确列出起始日期和结束日期 → 计算间隔天数 → 与法定期限比较,禁止跳步
- 审查判定完全依据steps中的审查方法,法律依据仅用于输出引用
- 判定结论必须是以下之一:"不存在问题"、"存在问题"、"缺少文书材料"
- 无法判断时标记"待人工复核"并说明原因
- 引用文书时使用文书名称,同一类型有多份时需区分
每个审查点的输出格式:
{
"review_point": "审查点名称",
"status": "不存在问题|存在问题|缺少文书材料|待人工复核",
"description": "问题描述(不存在问题则填'无')",
"recommendation": "处理建议(不存在问题则填'无')",
"referenced_documents": ["引用的文书名称列表"],
"legal_basis": ["法律依据列表"],
"reasoning": "分步推理过程(涉及时间计算时必填)"
}
阶段4:报告生成
- 使用 read_file 读取
Skill 资源目录/references/report-template.md获取报告模板 - 按模板格式,汇总所有审查点结论
- 输出完整审查报告
关键约束
- 法律依据仅用于输出引用,不参与审查判定。审查判定完全依据 steps 中的审查方法,legal_basis 仅在结论生成后查找并附上。禁止用法律依据的存无来影响判定结果。
- 分步推理:涉及日期或时间间隔计算时,必须分步推理——明确列出起始日期和结束日期→计算间隔→与法定期限比较,禁止跳步。
- 文书引用:引用文书时直接使用文书名称(如"行政案件立案登记表"、"张三的询问笔录"),同一类型有多份时需区分。
- 判定三值:每个审查点结论必须是以下之一:"不存在问题"、"存在问题"、"缺少文书材料"。
- 待人工复核:AI执行某步骤时无法判断时,标记"待人工复核"并说明原因。
- 文书缺失处理:required: true 的文书缺失→判定"缺少文书材料";required: false 的文书缺失→跳过依赖该文书的步骤,在报告中注明。
- 逐点审查:每个审查点独立审查,不因前一个审查点的结论影响后续审查点的判定。
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.
- 12d ago First seen · 119 lines · 45 tokens per session scan A 54428b3b0ae9
admin-review is a skill published in the GitHub repository ThomasMoreAI/legal-skills-open (72 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 45 tokens to every session and 1,707 once invoked, about $0.0002 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
pia-generator
Generate structured privacy impact assessments so Claude can organize project privacy risks, controls, and decision points into a review-ready report.
audit-scope-checklist
Generate a scoped IT audit program with controls mapped to roles and compliance frameworks. Use this skill for audit planning, control checklist generation, and auditprogram.json creation.
compliance-checker
Compare system configuration evidence against common control frameworks and produce a concise compliance assessment with gaps and remediation priorities.
compliance-controls-playbook
Guide Claude through a structured compliance audit workflow covering framework selection, evidence planning, control testing, exception handling, and reporting.
evidence-tracker
Manage audit evidence requests, collection progress, and status reporting so Claude can keep audits organized and identify overdue or blocked evidence items.
consent-checker
Review privacy policy or notice text against common privacy frameworks so Claude can explain coverage gaps, strengths, and compliance priorities.