Tools

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

• Third-party content exposure detected (high risk: 0.85). The required workflow runs curl -s https://ethskills.com/tools/SKILL.md at install/setup time, so the agent ingests outsider-authored free text fetched from a public web page into its LLM context.

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

• Potentially malicious external URL detected (high risk: 1.00). The catalog explicitly instructs the agent to fetch and read the remote file at https://ethskills.com/tools/SKILL.md at runtime (e.g., "curl -s https://ethskills.com/tools/SKILL.md" and "Tell your agent: 'Read https://ethskills.com/tools/SKILL.md'"), so external content can directly control agent prompts/instructions.