review-staged

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt requires loading and passing file contents (including ~/.claude/rules/* and repo files) verbatim to reviewers and then emitting diff/file snippets and file:line evidence in the report, which forces the model to read and potentially output any secrets (API keys, tokens, passwords, cookies) present in those files.