The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the atmos secret command family to perform administrative tasks including initialization, setting, and retrieving secrets. It also includes commands like atmos secret exec and atmos secret shell which inject sensitive values into the execution environment of other processes.
[DATA_EXFILTRATION]: The skill handles sensitive information by design. It includes mechanisms for downloading secrets to local files (atmos secret pull) and retrieving them for inspection (atmos secret get). Potential risks are mitigated by documentation instructions to enable masking and repository policies against committing unencrypted secret files.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from external secret backends (e.g., Vault, 1Password, AWS SSM). While these represent external data sources that could theoretically contain malicious content, the usage is confined to the primary purpose of the tool and follows established secret management patterns.

atmos-secrets