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 kingselyjoe/legal-research-dsh --skill legal-researchgit clone --depth 1 https://github.com/kingselyjoe/legal-research-dshWrote 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/kingselyjoe/legal-research-dsh/legal-research)<a href="https://agentmods.dev/skills/kingselyjoe/legal-research-dsh/legal-research"><img src="https://agentmods.dev/badge/skills/kingselyjoe/legal-research-dsh/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/kingselyjoe/legal-research-dsh/legal-research"><img src="https://agentmods.dev/badge/skills/kingselyjoe/legal-research-dsh/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 | $0.00078 | $0.01184 |
| Opus 5.5 | $0.00031 | $0.00474 |
| Sonnet 5.5 | $0.00016 | $0.00237 |
| Haiku 4.5 | $0.00008 | $0.00118 |
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 1mo 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 — 84 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Legal Research DSH
把法律研究做成可复核的证据工程。模型负责研究判断，legal-research 命令负责项目状态、来源、论点、证据和审计。
开始任何工作前完整读取并遵守 references/interaction-protocol.md。它规定如何向用户解释当前步骤、处理结果和下一步；原始 JSON 只供内部读取。
不可破坏的规则
- 不凭记忆编造法条、案号、裁判要旨、页码、DOI、引文或效力状态。
- 事实、法律命题、学术观点和作者分析必须分开标记。
- 引用法律结论前回到可访问的原始来源；只有摘要时明确标为二手来源。
- 缺失内容保留
[待核验]或[待补：具体字段]，不得猜补。 - 法域、研究截止日期、目标体例和交付格式必须进入项目元数据。
- 输出是研究辅助材料，不冒充律师意见、司法结论或学术原创性证明。
开始或恢复
先运行：
legal-research diagnose --json
legal-research --workspace "{workspace}" guide --json
没有项目时，根据用户已经提供的信息直接初始化；只有核心命题或法域确实无法推断时才追问：
legal-research --workspace "{workspace}" init \
--title "{题目}" --jurisdiction "{法域}" \
--research-type law-review --citation-style "中引法" \
--thesis "{暂定核心命题}" --json
上面的命令在内部执行。面向用户先说明当前主阶段、这一步的目的和完成标准；执行后说明实际产出、对论文的作用和唯一的建议下一步。
当前阶段达到完成标准后，在内部运行 legal-research --workspace "{workspace}" stage {下一阶段} --json，再重新读取 guide --json。不得仅因调用过某个模块就把阶段标成完成。
研究类型仅选 legal-memo、law-review、empirical、literature-review。
路由
| 用户目标 | 模块 |
|---|---|
| 法条、案例、文献和数据库检索 | legal-research-sources |
| 北大法宝等授权法律数据库的MCP、CLI和批量导入 | legal-research-providers |
| 拆论点、找反例、建立证据地图 | legal-research-claims |
| 中引法、Bluebook、OSCOLA等引注核验 | legal-research-citations |
| 论文计划、文献综述、备忘录和正文 | legal-research-writing |
| 降低模板化表达、强化作者判断和论文作者性审查 | legal-research-authorship |
| 投稿前审查、论证和方法复核 | legal-research-review |
| 裁判文书批处理、编码和实证数据 | legal-research-empirical |
| Markdown、DOCX和交付检查 | legal-research-export |
当环境不能直接激活同名 Skill 时，完整读取对应目录的 SKILL.md 后执行。项目目录规范见 references/workspace.md。
默认工作流
- 阶段 1/6 研究设定：固定问题、法域、时间边界、研究类型、核心命题和反命题。
- 阶段 2/6 来源检索与核验：需要授权数据库时先执行
legal-research-providers，再执行legal-research-sources登记并核验来源。 - 阶段 3/6 论点与证据链：执行
legal-research-claims建立证据地图，再由legal-research-citations核验身份、定位和体例。 - 阶段 4/6 提纲与正文：执行
legal-research-writing，正文只使用证据台账中可追溯的材料。 - 阶段 5/6 作者确认与审查：依次执行
legal-research-authorship和legal-research-review，修复阻断问题。 - 阶段 6/6 导出与交付：运行
legal-research audit --strict；通过后执行legal-research-export并人工核对。
What ships with it
3 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.
- 1mo ago First seen · 84 lines · 78 tokens per session scan A bd55584aa305
legal-research is a skill published in the GitHub repository kingselyjoe/legal-research-dsh (2 stars, last pushed 2mo ago), licensed MIT. It adds 78 tokens to every session and 1,184 once invoked, about $0.0003 per session on Opus 5.5. 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
deslop
Audits and rewrites legal writing, professional documentation, and general text to follow plain English principles. Translates archaic legalese and eliminates verbose "AI slop" by enforcing active voice, subject-verb proximity, and concise phrasing (inspired by Bryan Garner, the SEC Plain English Handbook, Federal…
xiaohongshu-prohibited-word
A Xiaohongshu content checker that finds platform-prohibited or sensitive words in text, documents, images, and web pages. Xiaohongshu is a Chinese social platform with rules that can limit or reject posts containing certain terms.
wechat-prohibited-word
A compliance checker for WeChat public-account content. It scans text, files, images, or web pages for restricted or sensitive wording and suggests safer replacements.
format-monitoring-report
A formatting guide for turning already-approved legal-monitoring content into a DOCX or PDF report. It specifies a seven-section layout, blue labels, Times New Roman, footnotes, page numbers, and a final comparison table.
verifiable-legal-document
A Russian-language workflow for producing legal documents with claim identifiers, linked supporting passages, hashes, and validation reports.
matter-intake
A setup process for a new legal matter or research assignment. It turns an unstructured request into a defined question, scope, data classification, plan, sources, deadline, and completion criteria.