contribute-to-eliza

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow invokes scripts/live-report.mjs which paginates and parses attacker-authored GitHub issue/PR bodies, comments, reviews, and evidence rows via gh api (e.g., issues list + comments + PR reviews), exposing that free text to the LLM for candidate selection and evidence auditing.

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

• Potentially malicious external URL detected (high risk: 1.00). The SKILL.md explicitly instructs the URL-only mission to read https://eliza.army/skill-manifest.json and fetch/compare https://eliza.army/skill.md at runtime to resolve a skill revision, meaning remote content can directly control the agent's skill/instructions.