1password
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches and installs the official 1Password CLI formula using the Homebrew package manager.
- [DYNAMIC_EXECUTION]: Uses the shell
evalcommand to apply environment variable exports generated by the 1Passwordop signintool. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external 1Password vaults, which could potentially house malicious instructions designed to influence the agent's behavior.
- Ingestion points: The agent reads secrets using
op readandop inject(specified inSKILL.mdandreferences/cli-examples.md). - Boundary markers: Includes explicit instructions in
SKILL.mdto prevent secrets from being pasted into logs, chat, or code output. - Capability inventory: The skill invokes shell commands via
opandtmux, and provides examples for writing data to local files (specified inreferences/cli-examples.md). - Sanitization: Recommends using
op runorop injectfor memory-based secret handling to minimize the need for writing sensitive information to disk. - [COMMAND_EXECUTION]: Executes system commands to interact with the 1Password CLI and manage persistent tmux sessions for authentication state.
Audit Metadata