custom-prompt

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.90). The file contains explicit instructions that would bypass or override user confirmation and system-mode protections (e.g., "Ignore Claude-specific mode-switch..." and "AUTO-SELECT: ... Do not ask the user..."), which are deceptive relative to the skill's stated safe behavior of mandatory confirmation and gating.

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill instructs the agent to solicit and resolve declared "inputs" from the user and then execute the prompt body (substituting those inputs into steps), which creates a high risk that user-supplied secrets (API keys, tokens, passwords) would be requested and then embedded verbatim into generated commands or code.