code-investigator

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow instructs the agent to “Read problem, logs, stack trace, test output, or bug description” (SKILL.md:23-23), which in typical runtime use is user/outsider-provided free text such as bug reports/log excerpts; the skill file itself doesn’t constrain or sanitize that input, so outsider-authored text can plausibly be ingested into the LLM context.