Loading ToolBench
Single tool with adequate schema and basic description. Tool name follows verb_noun pattern (get_gold_price). Input schema is complete with types and descriptions for all parameters. However, description is minimal (56 chars, below baseline 194 chars), lacks context on when to use vs alternatives, and does not document output structure. No error handling guidance. Parameters lack constraints (currency/metal should be enums). Output is raw JSON dump with no field documentation. Missing idempotency guarantees and rate-limit guidance for external API calls.
Get current gold price in specified currency
Tool description is too brief (56 chars vs baseline 194 chars). Lacks context on when to use, prerequisites (API key requirement), and output structure. LLMs cannot determine if this is the right tool without more detail.
Parameters 'currency' and 'metal' accept free-form strings but should be enums. 'currency' should enumerate valid ISO 4217 codes; 'metal' should enumerate XAU, XAG, XPT, XPD. Free-form strings invite hallucinated values.
Output schema is not documented. Tool returns raw JSON from goldapi.io with no field descriptions. LLMs cannot plan downstream operations or extract relevant fields without knowing the response structure.
Error handling is generic ('Gold API error: {str(e)}'). Does not guide LLM on recovery: is the error retryable? Should the user check the API key? Is the date format invalid? No actionable guidance.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | <=2025-11-25 | v2 |
API key is exposed as environment variable without validation or rate-limit handling. Tool makes external HTTP calls without timeout, retry logic, or rate-limit awareness. Runaway agent could overwhelm goldapi.io.