send-it

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill includes instructions that direct uploading UI screenshots to a public third‑party host and contains code that can copy local credential files into an isolated test home, which could enable unintended exposure of sensitive data (suspicious steering without clearly malicious intent).

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/autoreview, the workflow ingests free text originating from the PR topic branch by reading changed repository files (e.g., diff/base snapshots and file contents) and passes that content into reviewer subprocess/LLM prompts without any “specific item selection” step beyond being in the changed diff.