azure-batch

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt includes multiple examples that embed secrets verbatim (e.g., SAS tokens, storage account keys, BATCH_ACCOUNT_KEY, instrumentation keys, and accountKey fields) in CLI, JSON, and code samples, which would require an agent to output actual secret values if those placeholders are filled.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). The links are mostly Azure Batch service endpoints and Azure Blob/File storage URLs (including a direct .py script), and while the azure.github.io BatchExplorer page is official and low-risk, user-controlled blob/file storage hosting executable scripts or resource files is a common vector for distributing malware if the storage accounts are untrusted—so treat these as potentially suspicious.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The job manager task includes a resource file URL (https://mystorageaccount.blob.core.windows.net/scripts/submit_tasks.py) that is downloaded at runtime and then executed via "python3 submit_tasks.py", so external content fetched during skill runtime directly executes code and is a required dependency.