review-deep

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The orchestrator is required to call attach_packed_output, capture the returned outputId, and then include that outputId verbatim in subagent prompts and tool calls—forcing the LLM to handle and emit a sensitive access identifier in its outputs.

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

• Hidden Unicode characters detected (2 type(s) found)