Loading ToolBench
Server has 4 tools with complete input schemas and descriptions. Naming follows verb_noun convention (get_, list_, create_). Descriptions are present but brief (avg 60 chars, below 194-char baseline). All parameters have types and descriptions. No output schemas documented. Error handling is minimal, tool.handle() catches errors but returns generic JSON without recovery guidance. Missing tool annotations (readOnlyHint, destructiveHint). create_payment_link is the only write operation but lacks confirmation/dry-run pattern.
Crea un link de pago de Wompi listo para compartir.
Obtiene el token de aceptación requerido para procesar transacciones.
Consulta el estado de una transacción de Wompi por su ID.
Lista transacciones recientes en Wompi.
Output schemas not documented. LLMs cannot infer what fields create_payment_link or list_transactions return, forcing them to guess downstream field names for chaining.
create_payment_link (write operation) lacks confirmation/dry-run pattern. Agents can accidentally create payment links without user approval.
Tool descriptions are brief (avg 60 chars vs 194-char baseline). get_acceptance_token description (50 chars) lacks context on when/why to call it or what it returns.
Error handling returns generic JSON error messages without recovery guidance. 'Error: <message>' tells LLM nothing about retry eligibility or next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
Tool annotations missing. create_payment_link should declare destructiveHint=true; read-only tools should declare readOnlyHint=true for agent planning.