research

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill explicitly directs a background agent to publish all research findings to GitHub issues (and forbids local storage), creating an explicit, deliberate channel that can exfiltrate sensitive information to an external service if the agent collects internal or sensitive data.

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

• Third-party content exposure detected (high risk: 0.70). The required workflow launches a background agent that investigates “primary sources” and then writes findings to a GitHub issue, which implies runtime ingestion of outsider-authored free text from external web/docs (e.g., fetched URLs or scraped content) into the agent’s LLM context during research.