loom-skill-enhancement

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The required runtime path uses transition.inspect_existing_skill_markdown / transition.inspect_existing_package_lock / transition.inspect_existing_workflow_assets with stepKind: memoryRead to load checked-in files from the user-supplied target_skill_path (i.e., outsider-authored SKILL.md and assets/so-workflow/* content) into the agent/LLM context for analysis, which is an indirect prompt-injection risk.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs runtime download/probing of .nupkg artifacts (for example https://www.nuget.org/api/v2/package/Techne.Loom.SkillOrchestrator/0.2.151 and the beta example https://www.nuget.org/api/v2/package/Techne.Loom.SkillOrchestrator/0.2.148-beta) which are fetched during execution to extract and run so.dll, so the fetched remote content is a required runtime dependency that executes code and thereby controls the agent runtime.