1Password Cli For Agents

ReviewAudited by ClawScan on May 10, 2026.

Overview

This is a coherent instruction-only 1Password CLI helper, but it can let an agent read, change, delete, and possibly cache secrets, so use a tightly limited vault and token.

Install only if you are comfortable letting an agent use a 1Password service account. Create a dedicated vault, grant minimum permissions, keep OP_SERVICE_ACCOUNT_TOKEN private, require confirmation for writes or deletes, and avoid caching retrieved secrets.

Findings (3)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

If the token is too broadly scoped, the agent could access or change more secrets than intended.

Why it was flagged

The skill requires a 1Password service account token that can access vault data. This is expected for the integration, but it grants meaningful account authority.

Skill content
Create a Service Account and grant it access to a specific vault ... Set `OP_SERVICE_ACCOUNT_TOKEN` in your environment ... All commands require the `OP_SERVICE_ACCOUNT_TOKEN` to be set.
Recommendation

Use a dedicated 1Password vault, grant the service account only the minimum needed permissions, keep the token out of logs and source control, and rotate it if exposed.

What this means

A mistaken or overbroad command could modify or remove important secrets.

Why it was flagged

The documented CLI operations can create, edit, and delete 1Password vault items. This matches the stated purpose, but these are high-impact operations.

Skill content
op item create ... op item edit "Item Name" password="newpassword" ... op item delete "Item Name" --vault "Vault Name"
Recommendation

Require explicit user confirmation before create, edit, or delete operations, always specify the vault, and prefer read-only service account permissions when write access is not required.

What this means

Secrets could remain in an agent cache, transcript, or other temporary storage longer than the user expects.

Why it was flagged

The instruction to cache results is underspecified for a secrets workflow; cached results could include retrieved secret values.

Skill content
Rate Limits: Service accounts have rate limits. Cache results if possible or retry with backoff.
Recommendation

Do not cache secret values. If caching is necessary, cache only non-sensitive metadata, keep it short-lived, and clear it after the task.