Loading ToolBench
This server has critical deficiencies in naming, parameter documentation, and schema completeness. The 'vehicle_api' tool has an empty input schema with no parameters documented. The 'auth' tool has a moderately clear name and schema but lacks comprehensive parameter descriptions. Neither tool has descriptions that meet production quality standards. Error handling exists but is minimal. The codebase shows basic structure but fails to follow 54 core agentic patterns around parameter clarity, output documentation, and recovery guidance.
authenticate and manage authentication tokens
get vehicles List from API
vehicle_api has an empty input schema ({}), making it impossible for LLMs to understand what parameters it accepts or requires. Per hard scoring rule: if a tool has NO input schema at all, schema score MUST be 0.
vehicle_api tool name does not follow verb_noun convention. 'vehicle_api' is a generic technical name, not an action. Should be 'list_vehicles', 'get_vehicles', or 'search_vehicles' to clearly signal the action.
vehicle_api description is only 26 characters ('get vehicles List from API'). This violates the minimum 10-1024 character requirement and fails to explain WHEN to use it, WHAT it returns, or any prerequisites. Per hard scoring: descriptions under 20 chars score 0-40.
auth tool description is only 52 characters and does not clarify the three distinct actions (set, check, clear) or their preconditions. LLMs cannot infer that 'set' requires a token while 'check' and 'clear' do not.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 35 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 0 | - | v1 |
No output schema is documented for either tool. LLMs cannot predict what fields to expect or how to chain tool results downstream. Production tools must document return types with field descriptions.
vehicle_api tool implementation ignores authentication entirely, the requireAuth() call is commented out. This breaks the composition pattern: the auth tool cannot be used to gate the API call, defeating the security design.
No pagination support. vehicle_api calls swapi.py4e.com/api/vehicles/ without limit or offset parameters. If the endpoint returns 50+ vehicles, the response will bloat the context window. Per pattern: tools returning lists must accept page/offset and limit and return a count or next_cursor.
Error messages in auth tool lack actionable recovery guidance. 'Error: Token is required for set action' does not tell the LLM how to obtain a token or what format is expected. Per pattern: errors must guide recovery.
Parameter 'token' in auth tool lacks a description explaining the expected format, length, or validity rules. LLMs cannot validate input without explicit constraints in the description.
No tool annotations present (readOnlyHint, destructiveHint, idempotentHint). The auth 'set' action is destructive but not marked as such. Per current spec (2026-07-28): tool annotations help agents reason about side effects.