The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts (lakewatch_api_client.py and manager_api_client.py) use subprocess.run to invoke a local security component binary (CryptoAPI) on Linux for password encryption and decryption. This is a standard vendor-specific implementation for managing service account credentials and does not involve arbitrary shell execution or command injection from user input.
[CREDENTIALS_UNSAFE]: The skill follows secure credential management practices. Plaintext passwords are never stored; the configuration files (lakewatch_api_config.yaml and manager_api_config.yaml) only contain placeholders for encrypted_password. The encryption process is interactive and ensures passwords do not appear in process lists or logs.
[DATA_EXFILTRATION]: Network operations are restricted to the configured Huawei Cloud MRS Manager and LakeWatch endpoints. The skill performs read-only diagnostic queries (e.g., querying audit configurations or log keywords) and does not possess mechanisms to exfiltrate data to unauthorized third-party domains.
[PROMPT_INJECTION]: The SKILL.md and alarm knowledge base files (alarms/*.md) contain strictly formatted technical instructions and constraints. No attempts to override safety filters, bypass guidelines, or extract system prompts were detected.
[STATIC_ANALYSIS_REVIW]: Static analysis flags regarding 'reverse shell patterns' in 12001.md were determined to be false positives. These files contain diagnostic steps for troubleshooting SFTP connectivity and database issues (e.g., connection exceptions and port tests) which use technical terminology that can trigger heuristic detectors.

huawei-cloud-mrs-host-alarm-diagnose