Loading ToolBench
Model Context Protocol server for GERP (Google ERP) - exposes tools for order creation, audit viewing, and system status checks via JSON-RPC over STDIO
Three tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. gerp_status has empty input schema ({}), gerp_create_order has well-structured input but no output schema documented, gerp_audit_view has minimal parameter description. No tool descriptions exceed 100 chars or provide WHEN/WHY guidance. No error handling patterns visible. Tool names follow verb_noun convention but descriptions lack actionable context for LLM selection.
Gives the AI direct read-only access to the physical Compliance bounds via Cloud Spanner
Receives JSON parameters, structures them, and triggers the Saga via GraphQL mutation
Securely checks the bound infrastructure for the AI system
gerp_status has empty input schema ({}). No parameters documented; LLM cannot understand what inputs are accepted or required.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., transaction IDs, audit results).
Tool descriptions are under 100 chars and lack WHEN/WHY context. 'Securely checks the bound infrastructure' does not explain when to call gerp_status vs other tools or what 'bound infrastructure' means.
gerp_audit_view parameter 'target_record_id' has minimal description ('The record ID to audit'). No guidance on format, valid values, or what 'audit' returns.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
No error handling patterns visible. No recovery guidance, retryability classification, or actionable error messages documented.