The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts (get_dlo_schema.py and get_dmo_schema.py) utilize subprocess.run to call the Salesforce CLI (sf). This execution is limited to retrieving organizational metadata and authentication tokens for the specified target org. The command arguments are passed as a list rather than a shell string, effectively preventing command injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The documentation includes instructions to install standard Python libraries (requests and pyyaml) using pip. These are reputable, well-known dependencies required for making HTTP requests and handling data, posing no unusual risk.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of DLO and DMO schema metadata (labels, descriptions, and field types) retrieved from the Salesforce API. The agent is instructed to format and present this information. While any ingestion of external data constitutes a theoretical surface for indirect prompt injection, the risk is negligible as the data consists of structured metadata from a trusted enterprise platform.
[DATA_EXPOSURE]: The scripts handle Salesforce access tokens retrieved from the local environment via the CLI. These tokens are only used to authenticate requests to the designated Salesforce instance URL and are not stored, logged to disk, or transmitted to any third-party services.

data360-schema-get