workflow-creator

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md:66-67 requires using WebSearch / browsing https://www.skills.sh at runtime to discover skills, which can fetch public web content and feed that outsider-authored free text into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs downloading skills from the external registry https://www.skills.sh and lists external skill sources (mattpocock/skills, obra/superpowers, anthropics/knowledge-work-plugins, kepano/obsidian-skills) in manifest.json, which would be fetched at runtime and can directly control agent prompts/behavior.