Employment Law Research

US Employment Law Research

This skill produces a structured research brief on a specific area of US employment law. The note is the input for downstream work like building a dashboard, writing a memo, or briefing a stakeholder. The most important thing the skill does is enforce a clean separation between primary-source claims (the statute says X) and secondary-source claims (Cooley wrote about it last week).

Assume the audience is practicing lawyers.

Use this skill standalone when the user just wants research, OR as the first step before the employment-law-dashboard skill. If a dashboard is the eventual deliverable, complete this skill's workflow first and save the research note before invoking the dashboard skill.

When to use

The skill should fire whenever a user asks for research, comparison, or landscape on a US employment law topic. Examples:

• "Research the state of US laws on workplace surveillance."
• "What are the pay transparency laws across the US?"
• "Help me understand non-compete enforcement state by state."
• "I need a brief on pregnancy accommodation requirements nationally."
• "Make a dashboard for [employment law topic]" → run this first, then the dashboard skill.

Workflow

1. Clarify scope before researching

Before doing any web searches, lock down three things with the user:

• Topic — be specific. "AI in hiring" is too broad; "use of AI tools for employment decisions" is workable.
• Jurisdictional scope — federal layer + which states? Default to "federal + state regimes + laws from major cities (e.g. NYC, SF)", but confirm.
• Recency window — how recent does "recent" mean? Default last 12 months for legislation, last 6 months for reporting.

For dashboard prerequisites, you can usually inherit scope from the dashboard request directly without a separate clarification round.

2. Read the source policy

Before searching, read references/source-policies.md. This is the load-bearing constraint of the skill. It explains which kinds of statements need primary sources only, and which can use reputable secondary sources. Apply it as you research, not just at writeup time — if you find yourself reaching for a secondary source for a statutory claim, push back to the primary source.

3. Conduct research

Use available web search and fetch tools. Cover the layers in this order:

a. Federal layer

• Any federal statute, regulation, or controlling case law on the topic.
• Recent federal activity: executive orders, agency guidance, pending legislation, enforcement action.
• Primary sources for federal statutes: the U.S. Code on uscode.house.gov or gpo.gov. For regulations, the Code of Federal Regulations on ecfr.gov. For agency guidance, the agency's own .gov site.

b. State and city enacted laws

• Identify the most prominent enacted regimes (typically 3-5).
• For each: official statute citation, enacting bill number, effective date, enforcing agency, key obligations, key carveouts.
• Pull primary sources: state legislature websites, official rules databases, agency guidance pages on the agency's .gov domain. NYC laws are at codelibrary.amlegal.com/codes/newyorkcity and rules.cityofnewyork.us. State laws live on the state legislature site (e.g., ilga.gov, leg.colorado.gov, capitol.texas.gov).

c. Pending bills

• Approximate count of states with pending legislation.
• The 2-3 most-watched pending bills.
• Primary citation: the actual bill page on the state legislature site. Tracker citation (for the count) goes to a tracker org.

d. Litigation

Cover litigation at both the federal level and within the states. Only include cases that meaningfully shape the landscape. The bar: would a practitioner advising on this topic need to know about this case? Skip routine litigation.

Cases worth including:

• Cases that have enjoined, narrowed, or could enjoin a statute or regulation in scope (e.g., the FTC non-compete rule challenges).
• Cases that establish controlling precedent on a contested obligation.
• High-profile pending cases practitioners are tracking (e.g., Mobley v. Workday for AI hiring).
• Federal-level cases at any tier (district, circuit, Supreme Court) and federal agency adjudications (FTC, EEOC, NLRB).
• State-level cases that affect state statutes in scope.

For each case capture: case name, court, docket or citation, current status (pending, decided, on appeal, enjoined, settled), and a one-sentence framing of what it does to the law.

Primary citations: link to the court's docket page, the official opinion (CourtListener, PACER, or the court's own site), or the agency adjudication page. Coverage of the case from a law firm alert is Bucket B and belongs under "Recent reporting," not here.

e. Multi-state trackers

• Identify 2-4 reputable trackers covering this topic.
• Strong candidates: Littler, Bryan Cave Leighton Paisner, K&L Gates, Multistate.ai, Bloomberg Law, NCSL (caveat — NCSL has been less reliably updated for AI legislation since 2024; check before citing).
• Capture: org name, tracker title, URL, update cadence, brief description of what makes the tracker useful.

f. Recent reporting

• 3-5 articles on enacted laws (within the last 6 months preferred).
• 1-3 articles on pending bills (within the last 3 months preferred).
• Capture: source name (firm or publication), date, title, URL, one-sentence "why this is worth reading" framing.

4. Apply the source policy at writeup time

For every statement in the research note, classify it as Bucket A (primary-source-only) or Bucket B (secondary-source-welcome) per references/source-policies.md. Reject any Bucket A claim that lacks a primary citation. If you can't find one, that's a signal that the claim is wrong, oversimplified, or you're confusing two regimes.

Common failure modes:

• Citing a law firm summary for a "the statute requires X" statement → fix by going to the statute.
• Citing the bill number when you mean the codified statute → for enacted laws, prefer the codified citation (e.g., "775 ILCS 5/2-102(L)") over the bill ("HB 3773").
• Linking to a state legislature's "bill page" when the public act / final text is what's load-bearing → link to the public act / final text.

5. Produce the research note

Save the output as a single markdown file. Use this exact structure:

# [Topic] — US Employment Law Research

**Scope:** [topic, jurisdictions covered, recency window]
**Summary:** [brief overview of the main findings]
**Compiled:** [date]

## Federal layer
[Statutes / regs / EOs / agency guidance / pending federal action]
[Primary citations]

## Enacted laws by jurisdiction
For each jurisdiction (in alpha order, or grouped state/city):

### [Jurisdiction name]
- **Law:** [Codified citation; bill number in parens if helpful]
- **Effective date:**
- **Covered employers:**
- **Key obligations:** [notice / audit / disparate impact / anti-discrimination / etc.]
- **Enforcement:** [agency, private right of action, penalties, cure period]
- **Citations:** [primary sources only]

## Pending bills
- **Approximate count:** [N states] (cite tracker)
- **Notable pending bills:**
  - [Bill] — [State] — [status] — [link]
  
## Litigation
For each case worth including (federal or state):

### [Case name]
- **Court:** [court level and jurisdiction; docket or citation]
- **Status:** [pending / decided / on appeal / enjoined / settled]
- **What it does:** [one-sentence framing of effect on a statute or obligation in scope]
- **Citations:** [court docket or official opinion link]

## Resources

### Multi-state trackers
- **[Org name]** — [tracker title] — [URL] — [update cadence] — [why useful]
- ... 2-4 entries

### Recent reporting

#### Enacted laws
- **[Source]** — [date] — [title] — [URL] — [one-sentence framing]
- ... 3-5 entries

#### Pending bills
- **[Source]** — [date] — [title] — [URL] — [one-sentence framing]
- ... 1-3 entries

## Open questions / things to watch
[Items the user may want to monitor; uncertainties; unresolved interpretations]

6. Verify

Before handing off, read references/verification.md and run the checklist against the research note. Fix anything that fails. The verification pass catches claims that slipped through without primary sourcing, stale tracker citations, bill-vs-statute slips, and superseded litigation.

7. Hand off

Save the file in the user's workspace folder so they can read it and refer back. If this research is feeding into a dashboard, point the user to the employment-law-dashboard skill as the next step and tell them where you saved the research note.

Notes on common pitfalls

• Don't conflate bill and statute. A state legislature page for HB ###### is fine for tracking pending bills, but for an enacted law the codified citation (e.g., "775 ILCS 5/2-102(L)") is what attorneys actually cite. Find both; lead with the statute.
• Watch for stale "draft" or "proposed" rules. Many states publish draft regulations months before final adoption. Always note the status (draft vs final) and the effective date.
• Federal preemption attempts ≠ federal preemption. A federal executive order or pending bill that *attempts* to preempt state law isn't actually preempting until courts say so. Be careful with framing.
• City laws matter. NYC LL 144 is a city law, not a state law, and it's been a major part of the AI hiring landscape. Don't filter to "state only" by default.