release-ai-plugins
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local PowerShell scripts and Node.js modules to automate the release workflow. This includes versioning manifests, updating changelogs, and calling git CLI tools to detect modified files within the working tree. All execution is confined to scripts provided within the skill's own directory.
- [INDIRECT_PROMPT_INJECTION]: The generate-skill-registry.mjs script parses metadata from SKILL.md files to generate a manifest. Every Cat 8 finding requires a review of the ingestion surface. Ingestion points: parseFrontmatter in generate-skill-registry.mjs reads local SKILL.md files. Boundary markers: none identified beyond standard YAML frontmatter markers. Capability inventory: the skill can write manifest files, marketplace JSONs, and changelogs, and execute git commands. Sanitization: the script uses a restrictive regex-based parser to extract only specific fields (name, description, version), which significantly reduces the risk of accidental instruction execution from the ingested data.
- [PRIVILEGE_ESCALATION]: Documentation examples suggest running scripts with -ExecutionPolicy Bypass. While this is a standard developer practice for executing local, unsigned scripts in PowerShell, it involves bypassing default security restrictions on script execution.
Audit Metadata