The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes CLI tools in tools/ and tools-python/ that interact with the 1Password service by executing the official op CLI. This is performed using secure methods like Bun.spawn and asyncio.create_subprocess_exec, which mitigate shell injection risks by passing arguments as arrays rather than shell strings.- [EXTERNAL_DOWNLOADS]: The skill relies on the official onepassword-sdk Python package and recommends using the official 1Password CLI. These are trusted resources and represent standard integration patterns for the service.- [SAFE]: The skill implements logic to parse local .env files, which constitutes a potential ingestion surface for indirect prompt injection. However, the implementation is safe as it avoids dynamic code evaluation and uses secure command execution patterns.
Ingestion points: Local .env files are parsed by the parseEnvFile function in tools/src/utils.ts and tools-python/src/op_env/utils.py.
Boundary markers: Absent.
Capability inventory: The skill provides full CRUD capabilities for 1Password items and secret resolution via the official 1Password API and CLI.
Sanitization: Input variables are parsed into discrete values and passed directly to SDK methods or subprocess arguments, ensuring they are not evaluated as shell commands.

1password