sql-injection-error

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). This prompt explicitly instructs extracting credentials/secrets (e.g., username:password dumps), saving them as evidence, and returning new credentials/tokens in summaries, which requires the LLM to output secret values verbatim.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The document provides explicit, step-by-step instructions for error-based SQL injection data exfiltration, credential harvesting, post-exploitation (file reads, command execution, credential reuse) and pivoting—behaviors that constitute deliberate malicious abuse and high-risk backdoor/activity.

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow is centered on crafting and sending SQLi payloads to a target and then reading the target’s returned database error messages; those error strings are outsider-authored free text from the remote application (public/third-party system) that the agent ingests into its LLM context.