Loading ToolBench
MCP Server for Google Ads, depends on Google Ads API.
Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. list_accessible_customers has an empty input schema (no parameters documented), while search has a well-structured input schema with all parameters typed and described. However, neither tool documents its output schema, which is critical for LLM planning. Descriptions are present but generic, search's description is verbose (194 chars) but lacks guidance on when to use it vs alternatives. No error handling guidance, no pagination support documented, and no tool annotations (readOnlyHint, idempotentHint) despite both being READ_ONLY operations.
Returns ids of customers directly accessible by the user authenticating the call.
Fetches data from the Google Ads API using the search method Args: customer_id: The id of the customer fields: The fields to fetch resource: The resource to return fields from conditions: List of conditions to filter the data, combined using AND clauses orderings: How the data is ordered limit: The maximum number of rows to return
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields (e.g., customer_id from list_accessible_customers, or field values from search). Both tools lack return type documentation.
list_accessible_customers has empty input schema ({}). While correct for a no-parameter tool, the description is too generic (22 chars: 'Returns ids of customers directly accessible by the user authenticating the call.'). Lacks context on when to call it or what structure it returns.
No tool annotations (readOnlyHint, idempotentHint) despite both tools being READ_ONLY and safe to retry. Agents cannot infer safety properties without explicit hints.
search tool accepts 'conditions' and 'orderings' as free-form string arrays with minimal guidance. No examples of valid GAQL syntax, no validation rules, no error recovery hints. LLMs will hallucinate invalid queries.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | <=2025-11-25 | v2 |
No pagination support documented. search tool accepts 'limit' but no offset/page/cursor parameter. Large result sets will blow context window; no guidance on handling truncation.