Freelancer Contract Review
Reviews contracts from the freelancer's perspective, flags common traps, scores fairness, and produces actionable negotiation scripts before you sign.
Skill itself is restrained: only reads user-supplied contracts (Read/WebFetch) and generates a Markdown report; no destructive operations; report includes an explicit legal disclaimer. Deductions: URL ingestion sends full contract text to an overseas LLM service with no sensitive-data-flow disclosure; no stated confirmation mechanism or rollback concept; publisher identity unverified.
Instructions are structured and self-consistent, with a clear three-step failure path for unreadable contracts (report error, request alternative, do not proceed), giving intelligible failure feedback. Deductions: no tests, no edge cases (very long contracts, non-English contracts, scanned PDFs), partial coverage of WebFetch failure paths; key paths not reproduced by execution, so static ceiling of 10 applies.
Trigger is precise (/legal freelancer <file>), audience and scenario are clear, and the 14-lens scoring framework sets coherent boundaries. Deductions: no declared non-fit ranges (non-US contracts, non-contractor agreements); framework is deeply US-bound (IRS 20-factor, state non-compete laws, FTC), limiting usefulness for Chinese users; no Chinese-language support; core function depends entirely on the Anthropic API, with mainland-China reachability unverified.
Documentation is well layered (trigger, phases, report template, follow-up command hints) with good progressive disclosure. Deductions: repository license metadata unknown; no versioning, changelog, maintenance ownership or update path; README is marketing-heavy ('60 seconds', Skool funnel); environment assumptions (Claude Code, Python, reportlab) live in the README rather than the skill file itself.
Output spec is highly concrete (fairness score, 20-item Bill of Rights, negotiation scripts, next steps); if followed, marginal value over manual review is evident. Deductions: static review cannot verify actual output quality or direct usability; the 0-100 weighting is author-defined with no calibration evidence; README cost/benefit figures ($300-500/hr etc.) are unsourced; static ceiling of 7 applies.
Everything is author-claimed methodology: no runnable tests, no third-party validation, no real output samples. Legal references (IRS factors, state lists) carry no sources, and the skill itself admits unverified currency ('check current status' re FTC non-compete ban).
- URL ingestion sends the full contract to an overseas LLM service; do not use with sensitive business or legal text without assessing the data flow.
- The framework is deeply US-bound (IRS classification test, state non-compete law); applicability to mainland China and non-US jurisdictions is limited and scores may mislead.
- The fairness score and 'Bill of Rights' are author-defined heuristics, not legal advice; always consult a licensed attorney before signing.
- Core function depends on the Anthropic API; mainland-China network reachability is unverified.
- Repository license is unlabeled; confirm terms before use.
What does this skill do, and when should you use it?
This skill is one of 14 bundled in the ai-legal-claude repository, located at skills/legal-freelancer/SKILL.md. Invoked via /legal freelancer <file>, it analyzes a contract from the freelancer/contractor's side across 14 review dimensions, including misclassification risk, IP ownership, payment terms, and restrictive covenants. Output includes a 0-100 fairness score, a traps checklist, a 20-item Freelancer Bill of Rights checklist, and a prioritized negotiation script. The report is generated as Markdown and carries an explicit disclaimer that it is not legal advice.
Accepts a contract as a file path, pasted text, or URL; identifies the hiring party and freelancer; analyzes every clause through 14 lenses (IRS 20-factor misclassification test, IP assignment, payment terms, kill fees, scope creep, revision limits, non-compete/non-solicit, confidentiality, liability and indemnification, portfolio rights, insurance, taxes, dispute resolution); produces a 0-100 fairness score with an A-F grade; flags 12 common freelancer traps; and writes a FREELANCER-REVIEW-[date].md report containing the Bill of Rights checklist, risk-tiered findings, and a ranked negotiation script.
- A designer or developer checks a new client contract for traps like unlimited revisions or Net-90 payment before signing
- A contractor assesses whether a relationship risks employee misclassification under the IRS 20-factor test
- A freelancer verifies IP clauses: whether IP transfers only on full payment and whether pre-existing tools and portfolio rights are protected
- A consultant prepares prioritized revision requests and replacement clause language before negotiations
- A remote contractor on cross-border work checks venue, currency, and insurance requirements for reasonableness
What are this skill's strengths and limitations?
- Concrete, actionable framework: IRS 20-factor table, payment-term thresholds, and red/yellow/green flags per clause with explicit criteria
- Structured, action-oriented output: fairness score, traps list, 20-item Bill of Rights checklist, and ranked negotiation script
- Covers the most common freelancer contract risks, including state-by-state non-compete enforceability context
- Explicitly disclaims legal advice and recommends attorney review before signing
- One member of a 14-skill collection installed together; no documented standalone install
- License unknown — commercial use and redistribution terms are unclear
- No test suite or documented platform validation; analysis quality depends on the underlying model
- PDF reports and /legal negotiate are separate skills in the collection; this skill only refers to them
- Jurisdictional scope is not specified; some content (IRS tests, state non-compete rules) assumes a US context
How do you install this skill?
Run the repository's one-line installer (installs the full collection: 14 skills, 5 agents, and PDF scripts): curl -fsSL https://raw.githubusercontent.com/zubair-trabzada/ai-legal-claude/main/install.sh | bash. Requires Claude Code with an active Anthropic API key. The repo does not document installing this single skill in isolation.
How do you use this skill?
In Claude Code, run /legal freelancer <file> where <file> is a contract file path, URL, or pasted contract text. If the contract is unreadable, the skill reports the error and asks for another format rather than proceeding. Afterward, follow the prompt to run /legal negotiate for counter-proposal language or /legal report-pdf for a PDF report (requires Python 3.8+ and reportlab).