vulnerability-chaining

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow ingests “findings come from all specialists” (outsider-authored issue/PR/wiki/discussion text or other team members’ reports) and then re-reads/assembles them into attack paths, so free-form outsider-authored prose can be fed into the agent’s LLM context via the findings-to-chain step.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)