research-start

/research-start

1. Load ~/.claude/plugins/config/claude-for-legal-zh/legal-clinic/CLAUDE.md → jurisdiction, practice area.
2. Use the following workflow.
3. Define the question specifically. Build the roadmap: statute starting points (unverified), areas of case law (not specific cases), secondary source materials, search keywords.
4. If the student has already uploaded existing research results: synthesize and identify gaps.
5. Output prominently up front a "leads not authority" heading. Everything is a starting point for the student to verify.

/legal-clinic:research-start "[省份]房屋租赁中以丧失居住条件为由的抗辩"

---

Research starting points: a roadmap, not specific research conclusions

Purpose

Legal research is a core component of clinical education. But the initial stage — figuring outwhat to research, finding the right statutes, understanding the framework — is often the most time-consuming and least educationally valuable part. Students spend hours finding starting points before they can do the actual research work.

This skill produces starting points: statutes to consult, areas of case law to investigate, and search keywords for PKULaw and Faxin.These are all unverified. All non-authoritative. All leads for the student to pursue in depth.

This is both a pedagogical safeguard and an ethical safeguard. Students still learn to research. They just start from a better place.

Load context

~/.claude/plugins/config/claude-for-legal-zh/legal-clinic/CLAUDE.md → jurisdiction (province), practice area.

Workflow

Step 0: Prioritize seed files

Before building the roadmap, read the clinic's own seed files. Uploaded by the supervising attorney at cold start (manuals, filing guides, local court rules, intake forms, sample case files, prior memos) — they are pre-reviewed, jurisdiction-specific, and worth more in the student's first 20 minutes of research than any database query.

1. Read ~/.claude/plugins/config/claude-for-legal-zh/legal-clinic/CLAUDE.md → ## 种子文件. Identify any items whose purpose or filename matches the research area.
2. For each match, present it at the top of the roadmap output as apriority seed filesblock. Name the file, explain why it is relevant to this specific question, and state what it may cover vs. where external research is still needed.
3. If no seed file matches the question, say so directly ("No clinic seed file matches this question — go straight to primary sources"). Do not fabricate matches.
4. If the clinic has a LIMITED DATA flag, add a line noting it.

The roadmap still covers statutes, areas of case law, secondary source materials, and search keywords — seed files are the first lead, not a replacement for the rest. But present them first, so the student starts from the supervising attorney's precedent starting point.

Step 1: Define the question

What is the research question? Be specific. Not "residential lease defenses" — but "defenses to eviction for rent arrears in [province] on the ground of loss of habitability, specifically whether a broken heater qualifies, and whether the tenant needs to give the landlord advance written notice."

If the question is too broad, narrow it with the student: "That is three research questions. Let's take them one at a time. Which one first?"

Step 2: Build the roadmap

Starting points for statutes: Listpossiblyrelevant statutes. Clearly state that these are possible, not confirmed.

Possibly relevant statutes(unverified — confirm current validity and applicability):

• Relevant provisions of the Contract Part/Property Part of the Civil Code — search for "lease contract" "repair obligation"
• [Province/City] Housing Lease Management Regulations — may define specific conditions (heating, water supply)
• [核实每条引注的现行效力和准确性——法条可能已被修订]

Case law areas to investigate: Not specific cases — these areareas. The student finds the cases.

Case law areas:

• Judgments of the [Province] High People's Court or Intermediate People's Court on lease contract disputes — search for typical cases that establish adjudication rules
• Cases on what conditions constitute loss of habitable conditions — specifically cases about heating
• Cases on procedural prerequisites — does the lessee need to give advance notice? Is it necessary to escrow the rent?
• Cases on remedies — offset against overdue rent, or separately claim damages?

Administrative regulatory sources: If applicable (especially in administrative disputes):

Administrative sources:

• [Administrative agency] regulations
• Administrative guidance opinions or policy manuals — usually more current than regulations

Secondary literature sources for building the framework: Where to get the framework before going deep into primary sources.

Secondary literature sources (for building the framework, not for citation):

• [Province] real estate law practice manual (check the clinic library)
• Relevant continuing legal education materials (CLE)
• If the specific issue is disputed, check articles in core law journals

Search keywords: For PKULaw, Faxin, or other search systems used by the clinic.

Suggested search terms to try:

• PKULaw:"居住条件" /s "供暖" & [省份]
• China Judgments Online:居住条件 AND 供暖 AND [省份]
• Faxin:租赁合同 解除 居住条件
• Refine based on returned results — these are starting queries

Step 3: Flag uncertainty

If the skill is uncertain whether a source is relevant or currently valid:

[不确定：[省份]对此是否有具体的省级法规vs.仅系司法实践中形成的裁判规则——检索会告诉你]

Uncertainty is clearly stated, not hidden.

No silent supplementation. This skill produces leads, not authoritative citations — by design, the student tracks down the citations themselves. But if a query to the configured search tool (PKULaw) returns few or no results for a specific rule or case, say so and stop. Do not fabricate citations from web searches or model knowledge to fill a thin result set without asking. Say: "[Tool]'s search returned [N] results. Coverage of [rule] appears thin. Options: (1) broaden the search terms, (2) try a different search tool, (3) search the web — results will be flagged [网页搜索 — 需核实] and should be verified against primary sources before relying on them, or (4) stop here and flag the gap to your supervising attorney. Which do you choose?" The supervising attorney decides whether to accept lower-confidence sources.

Source attribution. For each suggested citation, note where it came from:[北大法宝]、[元典]、[中国裁判文书网], or use the MCP tool name for citations retrieved from legal research connectors; flag web search citations with [网页搜索 — 需核实]; flag citations recalled from training data with [模型知识 — 需验证]; citations provided by the supervising lawyer or case file [用户提供]. Citations marked `needing verification carry a higher fabrication risk and should be checked first. Never strip or merge the tags—they tell students which leads are original search results and which are AI model guesses that need to be verified against primary sources.

Step 4: Synthesize the uploaded search results (if any)

If the student has already done some searching and uploaded it: read it, identify what is covered and what is missing.

Based on the search results you already have:

• You already have: [summary of covered content]
• Gaps: [content suggested by the above roadmap but not yet found by you]
• [需核实：你引用的案例——[名称]——通过北大法宝/法信核实其是否未被后续判决推翻或限缩]

Output

═══════════════════════════════════════════════════════════════════════
  检索路线图 — 线索，非权威
  以下无任何内容为经核实的引注。每条法条、每个案例领域、
  每个检索关键词都是你检索工作的起手点。你核实现行效力、
  适用性和准确性。你找到实际案例。
  如果以下某条最终被证明是错误的或过时的，这是预期之内 —
  这是一张在哪里找的地图，不是检索的替代品。
═══════════════════════════════════════════════════════════════════════

# 检索路线图：[问题]

**管辖地：** [省份] | **实践领域：** [领域]

## 优先阅读的种子文件

[按第0步。列出任何与该问题匹配的诊所种子文件，附一行"本文件可能涵盖什么"的说明。如无匹配："无诊所种子文件匹配本问题——直接进入一手来源。"]

## 法条起手点（未核实）

[列表附核实标记]

## 案例法领域待调查

[领域，非具体案例]

## 行政 / 监管来源

[如适用]

## 二次文献来源（用于建立框架，非引用）

[列表]

## 检索关键词

**北大法宝：** [查询]
**中国裁判文书网：** [查询]

## 不确定性标记

[路线图确实不确定的所有地方]

---

## 如何使用这份路线图

1. 从一个二次文献来源开始以获取框架
2. 找到并阅读一手法条——确认上述引注是现行有效的
3. 运行检索，找到典型案例
4. 依赖前通过北大法宝/法信核实所有内容
5. 一旦你掌握了规则，回来运行 `/memo` 来搭建你的分析框架

## 本路线图不做什么

- **不给你可以直接使用的引注。** 上述每条引注都是待核实的线索，不是可依赖的权威。
- **不替你做检索。** 你做检索。这份路线图让你更快到达起跑线。
- **不替代北大法宝。** 那些工具里有实际案例。这告诉你该往哪里指。

---

**引注核实——使用前必须完成。** 以上引注由 AI 模型生成，未经核实。在依赖任何案例、法条或规则——或将其纳入当事人工作——之前，请通过北大法宝、法信、中国裁判文书网或你诊所的检索平台核实准确性和现行有效状态。将未经核实的引注标记给你的指导老师。

What this skill does not do

• Provide authoritative citations. Clear, by design. Students verify every citation before use.
• Replace legal research. It accelerates the "where to start" phase; the research itself remains the student's work.
• Guarantee the roadmap is complete. This is a set of starting leads. Research may find sources the roadmap missed—that is fine, that is research.