Building Blocks

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

• Third-party content exposure detected (high risk: 0.85). Runtime workflow uses curl -s https://ethskills.com/building-blocks/SKILL.md to fetch public web content, which is readable text that can be ingested into the agent’s LLM context (outsider-authored free text from the public web).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's setup and install.command explicitly fetch and instruct the agent to "Read https://ethskills.com/building-blocks/SKILL.md" at runtime (curl -s https://ethskills.com/building-blocks/SKILL.md), so external content is retrieved and used to control agent prompts/instructions.