review-comments

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime LLM context is populated from the user-provided “review comments” hand-off text (free-form quoted comment blocks), which can be authored by outsiders (e.g., GitHub review comments copied from other participants); the validator script only parses/validates and does not sanitize or restrict the comment text before it would be used by the agent.

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

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