Missing Protections Finder (AI Legal Assistant)
Scans a contract for protections that industry standards say should be there but aren't, then delivers ready-to-insert clause language with urgency ratings.
The skill is pure prompt logic using only Read/WebFetch on user-specified files or URLs; no external writes, no credential access, no destructive defaults; a legal disclaimer is explicit and output file naming/location is predictable, giving a narrow risk surface. Deductions: unknown license metadata, unverified publisher, no boundary disclosure for sensitive contract data fetched via WebFetch, and no rollback/uninstall guidance within the skill itself.
Instructions are well-structured and self-consistent: trigger syntax, step-by-step flow, seven contract-type checklists, universal protections table, and a complete output template — happy path plausible. Deductions: no tests or reproducible samples, no failure feedback for missing files, unreachable URLs, or scanned/PDF inputs, and coverage-score math left to model judgment; static review cannot exceed 10.
Audience (freelancers, small businesses) and trigger command are clear, with checklist tailoring per contract type showing boundary awareness. Deductions: no declared non-fit ranges (jurisdictional differences, non-English contracts), no Chinese-language support, README marketing claims (60-second review) may not match actual capability, and mainland-China reachability via WebFetch is untested.
Documentation is readable with clear command mapping, output templates, and practical guidelines. Deductions: unknown license, no versioning/changelog/maintenance ownership, curl|bash installer pattern, and marketing content (Skool community, inflated metrics) mixed into the repo; hidden assumptions about repo-level scripts not disclosed in the skill.
The skill has a concrete goal (find missing protections and supply insertable clause language) with a complete, directly usable output template and practical calibration guidance (weak-clause notes, complexity-adjusted urgency). Deductions: no sample outputs or third-party validation of output quality, clause accuracy unverified; static review cannot exceed 7.
The skill file itself is primary auditable material and the README file tree corroborates structure. Deductions: all value claims (60-second review, risk statistics, scoring system) are author marketing with no tests, sample outputs, or independent reproduction paths, and statistics like 82% are unsourced.
- Output is informational only; suggested clauses must be reviewed by a licensed attorney before use.
- Supported jurisdictions and languages are undeclared; applicability to Chinese-language contracts and non-common-law jurisdictions is unverified.
- Sensitive data flow when fetching external contract URLs via WebFetch is undisclosed; use local files for confidential contracts.
- Repository license metadata is unknown; confirm licensing before commercial use.
- All effectiveness claims (60-second review, risk statistics) are unverified marketing content.
What does this skill do, and when should you use it?
This is one of 14 skills bundled in the zubair-trabzada/ai-legal-claude repository, triggered via the `/legal missing <file>` command. It reads a contract from a file path, pasted text, or URL, identifies the contract type, then checks it against a type-specific checklist (SaaS, employment, MSA, NDA, partnership, lease, procurement) plus 17 universal protections. Every gap is reported with a CRITICAL/IMPORTANT/RECOMMENDED urgency rating, risk level, plain-English explanation, a real-world failure scenario, and complete insertable clause text. Output is a Markdown report with a Protection Coverage Score and a prioritized pre-signing action list, wrapped in a legal disclaimer stating it is not legal advice.
Accepts a file path (via the Read tool), pasted inline text, or a URL (via WebFetch); extracts contract type, parties, effective date, and governing law; selects the checklist matching the identified contract type and checks it against roughly seven contract categories' must-have protections plus 17 universal ones; for each missing protection it produces a name, risk category, urgency rating (CRITICAL/IMPORTANT/RECOMMENDED mapped to HIGH/MEDIUM/LOW RISK), why-it-matters explanation, real-world scenario, and full clause language with section-number placeholders; calculates a Protection Coverage Score as a percentage; writes a MISSING-PROTECTIONS-[contract-name].md file; and flags protections that are present but weak in an 'Existing Protections' table.
- A freelancer reviewing a client contract before signing wants to confirm liability caps, termination notice, and IP assignment terms aren't missing
- A small business owner vetting a vendor agreement wants a quick check for missing acceptance criteria, insurance requirements, and audit rights
- A startup handed a SaaS contract wants to verify data ownership, breach notification, and source-code escrow provisions are present
- A founding team reviewing an NDA wants to confirm definition exclusions, compelled-disclosure carve-outs, and survival terms exist
- An entrepreneur evaluating a joint venture agreement wants to check exit mechanisms, deadlock resolution, and IP ownership of joint work
What are this skill's strengths and limitations?
- Outputs complete, ready-to-insert clause text rather than skeleton placeholders like '[insert cap amount here]'
- Three-tier urgency rating plus a prioritized action list makes triage straightforward before signing
- Checklists are tailored per contract type (SaaS, employment, MSA, NDA, partnership, lease, procurement) instead of one generic list
- Coverage score and 'present but weak' annotations give a full picture of the contract's current state
- Calibrates urgency to contract complexity and value — a two-page letter agreement isn't held to full MSA standards
- Free and open, versus the $1,500–$3,000 cost of a basic professional contract review cited in the README
- Not legal advice — suggested clauses still require review by a licensed attorney, per the built-in disclaimer
- The repository has no declared license (License: unknown), creating uncertainty around commercial reuse
- No test suite, benchmarks, or user validation; output quality depends entirely on the underlying model
- Checklists reflect common-law-style contracts; applicability to other jurisdictions is unverified
- Sparse metadata (no repo topics) and documentation is oriented toward the whole 14-skill collection rather than this sub-skill
- Coverage of specialized contract types beyond the built-in list (equity agreements, cross-border deals) is not documented
How do you install this skill?
Install the full collection (which includes this skill): curl -fsSL https://raw.githubusercontent.com/zubair-trabzada/ai-legal-claude/main/install.sh | bash. One command installs all 14 skills, 5 agents, and the PDF generation scripts. Requires Claude Code with an active Anthropic API key; Python 3.8+ and reportlab are needed only for PDF reports. Uninstall via curl -fsSL https://raw.githubusercontent.com/zubair-trabzada/ai-legal-claude/main/uninstall.sh | bash.
How do you use this skill?
In Claude Code, run /legal missing <file> where <file> is a local contract file path, pasted contract text, or a URL to a contract document. The skill auto-detects the contract type and writes a MISSING-PROTECTIONS-[contract-name].md report into the input file's directory (or the current working directory for pasted text). Minimal standalone installation steps for just this skill are not documented in the source.
How does this skill compare with similar options?
Within the same repo, the flagship /legal review command runs 5 parallel agents for a full review with a Contract Safety Score, while this skill is the focused 'missing protections' sub-skill; use /legal risks for per-clause severity scoring and /legal negotiate for counter-proposal language. The README also suggests pairing with the AI Marketing Suite and AI Sales Team from the same series, though they don't overlap functionally with this skill.