oss-forensics-orchestration

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

• Third-party content exposure detected (high risk: 0.85). Phase 2 spawns multiple outsider-source investigators (e.g., GH Archive/public GitHub API/Wayback/local git clone of a public repo) whose fetched free-text content is ingested into the agent workflow and then summarized/used by the orchestrator’s LLM context (e.g., via “Evidence summary: ” and subsequent hypothesis/report generation).