review-multi-perspective

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the system to provide "the full content" and "token definitions" verbatim to subagents and requires reviewers to "Quote the specific code", so any secrets embedded in the supplied code/context would be reproduced in reviewer outputs.

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow reads the “full content of the files under review” (and related WF specs/tests) and injects that verbatim into the LLM context as {code}/{context} for multiple subagents, so if the reviewed files originate from outsiders (e.g., PR/issue/wiki text), their free text can reach the model at runtime.