Loading ToolBench
Neutral MCP server for 1Password integrations with opaque-by-default secret handling.
38 tools with consistent verb_noun naming and complete input schemas. Descriptions present but often generic (avg ~80 chars, below 194-char baseline). All parameters typed and described. Output schemas not documented. Strong security patterns (acknowledgement strings for destructive ops, secret reveal gates) but error handling lacks recovery guidance. Composition is sound, tools are single-responsibility and chainable via IDs. Missing: output schema documentation, error classification, and LLM-optimized descriptions.
Create a new environment variable in an item
Delete an environment variable
Get a specific environment variable value
List environment variables in an item
Update an environment variable
Create a new group
Delete a group
Get details for a specific group
Output schemas not documented. LLMs cannot infer what fields to expect from tool responses, forcing them to guess at downstream field names and risk broken tool chains.
Descriptions are generic and brief (avg ~80 chars vs 194-char baseline). Many lack context on WHEN to use the tool or what it returns. E.g., 'Update a group' does not explain what fields can be updated or what the response contains.
Error handling lacks recovery guidance. Tools do not document what errors are retryable, user-fixable, or fatal. E.g., 'item_search' does not explain what happens if the vault ID is invalid or if no results match.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2025-06-18+ | v2 |
List all groups in the account
Update a group
Grant a group access to a vault
Revoke a group's access to a vault
Update a group's vault access permissions
Archive an item
Create a new item in a vault
Delete an item from a vault
Get metadata for an item without revealing secret values
Restore an archived item
Search for items in a vault by query string
Update an existing item
List allowlisted op CLI commands and scripts
Reload the op CLI script allowlists
Run an allowlisted op CLI command or script
Generate a random or memorable password
Read a password field from an item, with optional plaintext reveal
Get SDK capabilities and supported features
Reveal a secret value with explicit acknowledgement
Get the status of the unrestricted runner
Deactivate a user
Get details for a specific user
Add a user to a group
Remove a user from a group
List all users in the account
Update a user
List all vaults accessible to the authenticated user
Run an allowlisted workspace command via 1Password Connect
List workspace trust configuration
Reload workspace trust configuration
Acknowledgement parameters (acknowledgePlaintext, acknowledgeMutation, acknowledgePermanent) lack clear documentation of what exact string values are expected. LLMs may guess wrong and fail the call.
Pagination not documented for list tools. vault_list, group_list, user_list do not specify if results are paginated, what the limit is, or how to fetch additional pages.