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 code-lawyer/Legal-Agent-Skills --skill legal-researchgit clone --depth 1 https://github.com/code-lawyer/Legal-Agent-SkillsWrote 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/code-lawyer/legal-agent-skills/legal-research)<a href="https://agentmods.dev/skills/code-lawyer/legal-agent-skills/legal-research"><img src="https://agentmods.dev/badge/skills/code-lawyer/legal-agent-skills/legal-research/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/code-lawyer/legal-agent-skills/legal-research"><img src="https://agentmods.dev/badge/skills/code-lawyer/legal-agent-skills/legal-research.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 | $0.00144 | $0.01838 |
| Opus 5 | $0.00072 | $0.00919 |
| Sonnet 5 | $0.00029 | $0.00368 |
| Haiku 4.5 | $0.00014 | $0.00184 |
Grade A, and why
legal-research 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 — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
法律研究与案例检索分析
文件定位（先读）
本 skill 由多个文件组成。下文所有 references/… 路径均相对于本 SKILL.md 所在目录。开始前先确定该目录（你被指向或读取本文件时所在的位置），之后所有读取都基于它解析，不要用当前工作目录去找。
何时使用
- 研究类：做法律研究、出研究报告、论证某法律问题、查某争议点的法律依据（"能否解除""是否有效""如何认定""论证……"）。
- 检索类：检索类案、出案例检索报告、查某类纠纷的裁判规则/裁判倾向/同案同判分析。
- 复合类：研究问题需类案支撑，或用户要"研究 + 附类案"。
任务边界
本技能只做法律研究报告 + 类案检索报告。不做以下事项：
- 合同审查 / markup 合同 / 红线稿 → 请使用
contract-review - 合同起草、企业核验、尽调合规、续约提醒、流程图、文书翻译
- 按模板填充文书（非论证起草）→ 使用
document-fill
MCP 硬前置（开工前必做，不可跳）
本 skill 靠专业法律库 MCP（法条 + 案例）做可核验的研究/检索——没有可靠法律库就没有真研究/检索，拿模型记忆里的法条案例充数正是最危险的幻觉源。因此 MCP 接入是硬前置，不是可选加速器：
- 开工先探测：进入检索/分析前，先探测环境是否已接入法条/案例 MCP（详见
references/00-routing-intake.md的「MCP 前置门」）。 - 未接 → 阻断 + 引导：未探测到任何法律库 MCP 时，不产出正式研究/检索报告；先按
references/09-mcp-setup.md引导用户接入（申请本人 API Token → 填入宿主 agent 的mcpServers配置 → 重连），接入后再开工。 - 唯一例外（显式 opt-in）：用户被告知后仍明确要求"仅出初步参考"，才产出**通篇标注「未接法律库·仅初步参考·不可作为诉讼或交付依据」**的降级草稿；这不是默认行为。
执行顺序（按阶段按需读取，不一次读全）
- 先读
references/00-routing-intake.md— 判定模式（研究/检索/复合）、过 MCP 前置门、并建档。 - 所有模式都经
references/10-retrieval-core.md做检索核验（MCP 核验为主源；模型知识仅在 MCP 已接、个别命题查空时作降级标注，不是常规兜底）。 - 按命中模式读对应骨架：研究/复合走
20-research-skeleton.md，检索走21-case-skeleton.md。 - 分析中按触发条件加载
30-analysis-guardrails.md的对应护栏项，不预加载全部。 - 按
40-output-research.md/41-output-case.md成稿，过48-qc-gate.md终检门后交付。
硬不变量（不可跳过，常驻）
无论走哪条路径、读没读全深层文件，以下底线始终生效；48-qc-gate.md 终检门会逐项复核：
0. MCP 前置：未接法律库 MCP 时不产出正式报告（见上「MCP 硬前置」）。正式报告须建立在 MCP 核验来源之上，整份零 MCP 核验的报告不得作为正式交付。
- 来源标签：每条法条/案例命题必须挂且仅挂一个验证渠道标签（
[MCP核验:…]或[模型知识-未验证]）。格式见references/49-citation-disclaimer.md。 - 禁止编造：不得编造案号/条号；无来源标签的案号/条号一律删除或标未验证。
- 置信度门禁：仅凭模型知识的高风险命题（精确条号/阈值/最新司法解释状态）不得作为结论的唯一支撑。
- 终检门：报告产出前必过终检门（见
references/48-qc-gate.md），这是唯一不可跳的关卡；不通过的项按其精确回退目标退回对应步骤补正。 - 免责声明：任何正式交付必附完整免责声明（定稿见
references/49-citation-disclaimer.md），不得省略。
简单问答捷径
若任务是小问题（"某法条第几条说什么""某术语定义"），可在骨架内直接作答，无需加载全部分支：经 10-retrieval-core.md 核验层查一次、挂来源标签、直接回答；涉时效敏感点（法条修订、司法解释更新）仍按 10-retrieval-core.md 的强制复核触发条件处理。不因是"小问题"而跳过来源标签与核验状态。
What ships with it
13 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.
- README.md 6.5 KB
- references/00-routing-intake.md 11 KB
- references/09-mcp-setup.md 3.8 KB
- references/10-retrieval-core.md 11 KB
- references/20-research-skeleton.md 7.8 KB
- references/21-case-skeleton.md 11 KB
- references/30-analysis-guardrails.md 5.8 KB
- references/40-output-research.md 7.6 KB
- references/41-output-case.md 8.6 KB
- references/48-qc-gate.md 9.1 KB
- references/49-citation-disclaimer.md 7.1 KB
- tests/test_lr_validate.py 2.9 KB runs code
- validate.py 3.9 KB runs code
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 · 60 lines · 144 tokens per session scan A 8de3e2554fed
legal-research is a skill published in the GitHub repository code-lawyer/Legal-Agent-Skills (2 stars, last pushed 20d ago), licensed MIT. It adds 144 tokens to every session and 1,838 once invoked, about $0.0007 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 2026-08-31.
Other skills, from other repositories
contract-guardian
A contract-review workflow that checks agreements for risks, links important findings to relevant laws, and produces a written review. It covers common agreements such as sales, leases, loans, services, contracting, and intellectual property.
法律周报
A legal-news briefing workflow that finds recent court and legal-practice articles, scores them against selected criteria, and archives chosen items in a local Obsidian notes folder. Obsidian is a local note-taking application.
legal-research
A Chinese-language legal research workflow for finding laws, regulations, court cases, compensation standards, and legal grounds.
legal-citation-comprehensive
A Chinese and foreign legal-citation tool for checking, completing, explaining, and formatting footnotes under the second edition of a legal citation manual. It handles laws, cases, legal documents, and academic sources when the relevant source material is provided.
legal-fact-checker
A Chinese-language fact-checking guide for legal documents such as contracts, case analyses, legal opinions, and legal assignments. It checks provided drafts against factual and legal support without inventing laws, cases, or facts.
workflow-overview
A workflow guide for handling complex legal matters through assessment, issue identification, source research, reasoning, fact checking, and document output.