debug

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

• Third-party content exposure detected (high risk: 1.00). The workflow explicitly pulls public, user-generated GitHub issue content (via gh search issues / gh issue view --comments and writes it to /tmp/github.txt which is then included in prompts like $(cat /tmp/github.txt)), so the agent ingests untrusted third-party content that could contain injected instructions.