Loading ToolBench
Multi-service AI platform that integrates with GitHub, Notion, Slack, Gmail, Google Calendar, Salesforce, and web search via MCP servers. Provides a chat interface with OAuth-based service authentication and tool execution.
Critical gaps across all dimensions. Tool names lack clear action verbs (e.g., 'redirect_to_auth' is vague about what happens). Descriptions are present but generic (10-50 chars), failing to explain WHEN to use each tool or what distinguishes similar ones. Input schemas visible but lack depth: parameters have types and basic descriptions, but no enums for constrained values (e.g., 'service' param in redirect_to_auth accepts free-form strings instead of enum). No output schemas documented. Error handling absent, no guidance on recovery, retryability, or actionable error messages. Google Calendar tools lack idempotency guidance and confirmation patterns for destructive operations. Overall, definitions read as API pass-throughs rather than LLM-optimized tool interfaces.
Create a new event in Google Calendar.
Delete an event from Google Calendar.
List upcoming events from Google Calendar.
List all available external services that can be connected.
Triggers an automatic login redirect for the specified service. Supported services: github, notion, google, slack, salesforce
Vague action verb in 'redirect_to_auth', does not clearly signal authentication flow initiation. Should be 'initiate_service_auth' or 'authenticate_service'.
No enum constraints on 'service' parameter in redirect_to_auth. Accepts free-form strings instead of explicit enum [github, notion, google, slack, salesforce]. LLMs will hallucinate invalid service names.
Destructive tool (google_calendar_delete_event) lacks confirmation pattern or dry-run capability. No guidance on retryability or error recovery. Agents may delete events without safeguards.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., event_id from create_event for later deletion).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
Descriptions are too short (10-50 chars) and lack context. 'List upcoming events from Google Calendar' does not explain when to use this vs. other discovery tools, or what fields are returned.