ip-clause-review

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

• Third-party content exposure detected (high risk: 0.75). The skill ingests the agreement text provided by the user at runtime (file/Drive/pasted text) and then feeds it into the LLM for clause review; if that agreement text is authored by an outsider (e.g., a vendor/customer counterparty), it becomes outsider-authored free text in the LLM context via the “Get the agreement” step.

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

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