The Agent Skills Directory

[SAFE]: The skill installs the official 1Password CLI (op) via Homebrew, which is a well-known and trusted package management service.
[SAFE]: Employs tmux sessions as defined in SKILL.md to handle CLI authentication requirements and maintain session state securely. This prevents credential-related prompts from failing or leaking into the shell's command history.
[SAFE]: Includes explicit guardrails in SKILL.md that instruct the agent to avoid writing secrets to disk and to never output sensitive vault data into chat or logs.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests data from 1Password vaults which is then processed by the agent.
Ingestion points: Secrets retrieved from the 1Password vault via op read or template files used in op inject (evidence in references/cli-examples.md).
Boundary markers: There are no instructions to use delimiters or ignore instructions found within the retrieved secret content.
Capability inventory: The skill can execute commands with injected secrets via op run and write processed templates to the filesystem via op inject (evidence in references/cli-examples.md).
Sanitization: No sanitization or validation of the retrieved vault content is performed before the data is integrated into the agent's workflow.

1password