The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides several ingestion points for untrusted data that could potentially influence agent behavior or be used for malicious purposes. \n- Ingestion points: The cloud create command accepts natural language via --prompt and --prompt-file flags, and external code/data via the --repo flag. The cloud message command also accepts arbitrary text input from the user or external sources.\n- Boundary markers: There are no explicit delimiters or instructions provided to separate these inputs from the agent's core instructions.\n- Capability inventory: The skill utilizes the omniroute CLI to perform network operations (syncing to cloud providers) and sensitive file system operations (creating, managing, and restoring backups).\n- Sanitization: No sanitization, validation, or escaping of the ingested content is described in the provided skill instructions.\n- [COMMAND_EXECUTION]: The instructions direct the user to perform a global installation of an NPM package (omniroute) and execute various subcommands that interact with the system's files and network.

cli-backup-sync