Worker Classification

/worker-classification

Runs the applicable classification tests for the jurisdiction and flags where the proposed arrangement doesn't match the structure you're trying to use. Prospective only — for existing relationships, consult counsel.

Instructions

1. Load ~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.md → jurisdictional footprint, escalation table.
2. Run the full workflow below.
3. If the attorney provides details upfront, extract what's available and ask only about the gaps. Do not re-ask information already provided.

Examples

/employment-legal:worker-classification
We want to bring on a data scientist for 6 months, working out of our
SF office, using our tools, embedded in our analytics team.

/employment-legal:worker-classification
Is our recruiter contractor arrangement okay? She works exclusively for
us, sets her own hours, uses her own laptop, project fee per placement.

/employment-legal:worker-classification
(skill will ask for details)

Matter context

Matter context. Check ## Matter workspaces in the practice-level CLAUDE.md. If Enabled is ✗ (the default for in-house users), skip the rest of this paragraph — skills use practice-level context and the matter machinery is invisible. If enabled and there is no active matter, ask: "Which matter is this for? Run /employment-legal:matter-workspace switch <slug> or say practice-level." Load the active matter's matter.md for matter-specific context and overrides. Write outputs to the matter folder at ~/.claude/plugins/config/claude-for-legal/employment-legal/matters/<matter-slug>/. Never read another matter's files unless Cross-matter context is on.

Purpose

The most expensive classification decision is the one nobody made consciously. Someone describes what they want ("a contractor"), the engagement starts, and two years later the facts look like employment. This skill walks the applicable tests on the proposed arrangement before it starts — and tells you when what you're describing doesn't match the structure you're trying to use.

This skill teaches the reasoning pattern. It does not state the law. Every test formulation, statutory citation, threshold, and carve-out must come from current research for the applicable jurisdiction.

Prospective-only hard gate — run BEFORE intake

This skill analyzes a PROPOSED engagement before the work starts. Before any substantive intake (Step 1), ask:

Has this work already started? Is the worker currently engaged, or have they been performing work under this arrangement for any period of time (days, weeks, months, or years)?

If the answer is yes — the engagement already exists, in any form, for any duration — STOP. Do not proceed to Step 1 intake. Classifying an existing arrangement is not a planning exercise; it's a liability assessment with remediation implications: back pay (OT, meal/rest premiums), unpaid employer-side payroll tax, benefits eligibility that was denied, unemployment and workers' comp back-exposure, state penalties (in CA, PAGA), IRS § 530 relief analysis, and — in strict-test jurisdictions with ongoing work — the prospective exposure of letting it run another day. That analysis is privileged, led by counsel, and coupled with a remediation plan.

Output exactly this block and wait for a response:

Out of scope — existing arrangement.

This skill is designed to analyze a worker engagement before it starts, so the classification choice informs how to structure the contract and operations. You've described an arrangement that already exists. Analyzing an existing engagement retroactively is a different exercise: reclassification risk assessment coupled with remediation planning — back-pay exposure, payroll-tax back-exposure, penalty exposure, benefits exposure, IRS § 530 relief analysis, and prospective restructuring. That work should be privileged, led by an attorney, and likely coupled with outside-counsel review given the dollar and enforcement exposure.

Recommended next step: escalate per your config's escalation table (for retroactive classification, this typically routes to GC + outside employment counsel). I've flagged this for escalation routing.

If you want to proceed with the prospective-style analysis anyway for planning purposes, say "proceed anyway" — but understand:

• The output is NOT a remediation plan and should not be treated as one.
• The output does NOT scope back-pay, penalty, or payroll-tax exposure for the period already worked.
• The output does NOT substitute for the reclassification-risk assessment that this fact pattern actually calls for.
• The output will carry a prominent banner reflecting this scope mismatch, and the consequential-action gate will require an attorney yes before the analysis is treated as reliable.

Only say "proceed anyway" if you're using this skill for forward-looking planning (e.g., "if we were structuring this fresh today, how should we think about it?") and you have a separate plan for the remediation question.

Only proceed past this gate with an explicit "proceed anyway" (or equivalent user instruction). A hesitant "I guess" does not count — re-prompt. If the user proceeds anyway, prepend this banner to every output of this skill for this session:

⚠️ SCOPE MISMATCH — OUT-OF-SCOPE USE
This skill analyzes prospective worker engagements. The arrangement here
already exists. This output is the prospective-style analysis the user
requested for planning purposes only — it is NOT a remediation plan, does
NOT scope existing back-pay / penalty / payroll-tax exposure, and does
NOT substitute for the reclassification-risk assessment this fact pattern
requires. The remediation question has been flagged for escalation to
counsel per your config's escalation table.

If the answer to "has this work already started?" is no (the engagement is genuinely prospective, not yet begun), proceed to load context.

Load context

Read ~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.md → jurisdictional footprint, any classification history or prior settlements noted, escalation table, and any house classification policy the team has recorded.

Output header

Prepend the work-product header from ~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.md → ## Outputs (it differs by user role — see ## Who's using this).

Workflow

Step 1 — Information gathering

Ask all of the following in a single block. Do not drip questions one at a time. Briefly explain why you're asking — attorneys answer better when they understand what the question is testing.

To run the right classification tests I need to understand the proposed arrangement in detail. Please answer as many of these as you can — the more complete the picture, the more accurate the analysis:

The work

• What will this person actually do day-to-day?
• Is this work part of your company's core business, or peripheral to it? (e.g., a software engineer at a software company = core; an IT contractor at a law firm = more peripheral)
• Is this a defined project with a clear end, or ongoing indefinite work?
• How specialized is the skill? Does this person have expertise your team doesn't?

Control

• Who sets their hours and schedule — them or you?
• Where will they work — your office, their location, or either?
• Will you direct how they do the work (methods, process, sequence), or just what the end result should be?
• Will they supervise any of your employees?

Economics

• How will they be paid — hourly, daily, or fixed project fee?
• Will you provide equipment, tools, or software, or do they use their own?
• Do they work for other companies, or will this be exclusive?
• Will they bear any financial risk — can they profit beyond the fee, or lose money on the engagement?
• Do they have their own business entity (LLC, S-corp, sole proprietor)?

The arrangement

• How do you want to structure this — direct contractor, staffing agency temp, or vendor/SOW (company-to-company)?
• If staffing agency: who pays the worker — the agency or you? Who controls day-to-day work?
• Will there be a written contract? Do you have a template in mind?
• Roughly how long is the engagement — weeks, months, over a year?
• Will they work alongside your employees doing similar work?

Purpose(s) of the classification

• What legal purposes does the classification need to serve — federal payroll tax, FLSA wage/hour, state wage/hour, unemployment insurance, workers' compensation, benefits eligibility? Different purposes are often governed by different tests, and the answers can diverge.

Jurisdiction

• Where will this person physically perform the work?

Wait for responses before proceeding. If the attorney can't answer certain questions, note the gaps — they affect the analysis.

Step 2 — Identify the applicable tests

Research the applicable tests before proceeding. For the jurisdiction(s) and purpose(s) identified in intake, research the currently operative classification test(s). Jurisdictions commonly use one or more of: an ABC test, an economic-realities test, a common-law right-to-control test, a hybrid, or a purpose-specific statutory test. The test that governs for federal payroll tax may not be the same test that governs for state wage/hour, unemployment, or workers' compensation — run each purpose on its own track. Cite the controlling statute, regulation, or case. Note the effective date of each rule and whether it has been recently amended. Identify any carve-outs or exceptions that may apply (e.g., B2B, professional services, construction, referral-agency, business-to-business contracting relationship). Verify currency. If you are uncertain about the current state of the law in any jurisdiction, flag it for attorney verification — do not state a test you haven't confirmed.

If ~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.md records the company's house classification policy, apply it first and flag any tension with the researched test.

No silent supplement. If a research query to the configured legal research tool returns few or no results for a jurisdiction-and-purpose combination, report what was found and stop. Do NOT fill the gap from web search or model knowledge without asking. Say: "The search returned [N] results from [tool]. Coverage appears thin for [jurisdiction / purpose / test]. Options: (1) broaden the search query, (2) try a different research tool, (3) search the web — results will be tagged [web search — verify] and should be checked against a primary source before relying, or (4) flag as unverified and stop. Which would you like?" A lawyer decides whether to accept lower-confidence sources.

Source attribution. Tag every citation — each classification test, statute, regulation, or case — with where it came from: [Westlaw], [CourtListener], or the MCP tool name for citations retrieved from a legal research connector; [web search — verify] for web-search citations; [model knowledge — verify] for citations recalled from training data; [user provided] for citations the attorney supplied. Citations tagged verify carry higher fabrication risk and should be checked first. Never strip or collapse the tags.