azure-app-insights

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt's discovery commands explicitly retrieve instrumentation keys and mentions connection strings, which are secrets that the agent would need to fetch and could be output verbatim, creating a high exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime path uses az monitor app-insights query --analytics-query ... --output json to fetch KQL query results (including fields like outerMessage from exceptions), which are outsider-authored free text from application telemetry/logs; this text is ingested into the agent’s LLM context via the command’s JSON output.