issue-work-loop

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

• Third-party content exposure detected (high risk: 0.78). Yes: in the required runtime workflow, the skill queries GitHub issue/PR title/body and sends reviewer/implementer/fixer prompts that include those untrusted texts (e.g., PR title/body and any linked issue context) into the worker LLM context via herdr-agent-comms using the role prompts.