ecosystem-research

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow explicitly searches and ingests GitHub issue/discussion content (outsider-authored text) via Step 1 (“Search the primary repository's issue tracker and discussion board for real-world reports”), which would be read as free-form prose by the agent to produce summaries and citations.