ui-pack-maintain
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted UI pack data (Markdown, HTML, JSON) to generate registry metadata and facilitate semantic reviews. * Ingestion points: Pack directories provided via the
--packargument, specifically reading manifest.json, PACK.md, and component source files. * Boundary markers: The system utilizes YAML frontmatter delimiters (---) and structured JSON parsing to separate metadata from content. * Capability inventory: Includes file system read and write operations (readFile, writeFile, readdir) targeting the .html-prototype workspace and sister skill directories. * Sanitization: Employs regular expressions for frontmatter extraction and automated validation logic in validate-pack.mjs to check for forbidden patterns in script adapters. - [COMMAND_EXECUTION]: The skill includes Node.js utility scripts (validate-pack.mjs, resolve-pack.mjs, generate-registry.mjs) that perform automated file system tasks locally to maintain the integrity and discoverability of UI components.
Audit Metadata