Loading ToolBench
Server has 3 tools with significant quality gaps. Two 'weather' tools are duplicates with minimal descriptions (16-34 chars, below 20-char floor). The 'tridionSites' tool has a vague description (34 chars). All tools lack parameter descriptions, only type is specified. No output schemas documented. No error handling guidance. Descriptions are too short to guide LLM tool selection per pattern:tool-description baseline (194 chars avg). Tool naming lacks action verbs ('weather' and 'tridionSites' are nouns, not verb_noun pattern). Duplicate tool names ('weather' appears twice) violate composition pattern and confuse LLM routing.
Get the tridion sites for a city
Get weather for the city
Get current weather for the city
Duplicate tool names: 'weather' registered twice (netlify/functions/index.ts). LLM cannot distinguish between them; routing is ambiguous.
Tool names lack action verbs. 'weather' and 'tridionSites' are nouns. Should be 'get_weather' and 'get_tridion_sites' per verb_noun convention.
Descriptions too short (16 - 34 chars, below 194-char baseline). 'Get weather for the city' and 'Get the tridion sites for a city' lack context for LLM selection. No WHEN/WHY guidance.
Parameter 'city' has no description in any tool. LLM cannot infer whether to pass city name, code, or ID. Violates pattern:tool-description requirement.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
No output schemas documented. LLMs cannot plan downstream calls or extract fields. Response structure (temp, forecast) is implicit, not declared.