What problem does it solve? Legal teams installing community-built or first-party AI skills have no structured way to judge whether a skill is well-designed, safe, or trustworthy before it touches their workflows. This Skill provides a rigorous quality-assurance review that surfaces design gaps, security risks, and legal-specific failure modes before installation or deployment. ## Core Features & Use Cases - Thirteen-Parameter Design Evaluation: Scores audience, work shape, delegation threshold, input requirements, versioning, confidence bands, failure modes, scope boundaries, escalation logic, trust surface, freshness, schema, and conflicts with installed skills. - Prompt-Injection Heuristic Scan: Flags override instructions, authority claims, out-of-scope reads/writes, hidden content, external URLs, and credential asks, with fail-closed regression checks on skill updates. - Legal Failure Mode Check: Separately assesses legal advice vs. legal support, privilege implications, and accountability gaps, then issues a Ready / Some Concern / Material Concerns / Refuse verdict. - Use Case: Before installing a community NDA-review skill from a registry, run this QA to get a dependency map, injection scan results, a parameter-by-parameter gap table, and a clear verdict on whether to trust it. ## Quick Start Run /legal-builder-hub:skills-qa on the path to the skill directory you want to evaluate before installing it.
