The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/agent.py executes restic and aws cli commands to manage backups and S3 configurations. These are standard operations for the skill's stated purpose.
The implementation uses list-based subprocess.run calls, which prevents common shell injection vulnerabilities by not invoking a shell for command execution.
[CREDENTIALS_UNSAFE]: The skill demonstrates safe credential management practices.
It utilizes environment variables (RESTIC_PASSWORD, AWS_ACCESS_KEY_ID, etc.) or local files for sensitive information rather than hardcoding secrets within the code.
[DATA_EXFILTRATION]: Network operations are restricted to communication with S3-compatible storage providers (e.g., AWS S3) for backup purposes. No unauthorized data transmission to unknown third-party domains was detected.
[INDIRECT_PROMPT_INJECTION]: While the agent processes user-supplied file paths and repository URLs, it does not use these in a way that would allow for prompt injection or unauthorized code execution, as it relies on structured command lists for subprocess execution.

implementing-immutable-backup-with-restic