Loading ToolBench
6 tools with complete schemas and descriptions, but quality is uneven. Tool names follow verb_noun convention (generate_image, list_models, get_credits). Descriptions are present but brief (avg ~50 chars, below 194-char baseline). Parameters have types and descriptions, but lack constraints (enums, ranges, patterns). No output schemas documented. Error handling and recovery guidance absent. Security concern: upload_file accepts local paths without validation hints.
Generate images using Kie.ai models
Generate videos using Kie.ai models
Get current API credit balance
List available Kie.ai models with optional filtering
Upload a file to Kie.ai for use in generation tasks
Wait for a generation task to complete and retrieve results
Descriptions are too brief (avg 50 chars vs 194-char baseline). 'Generate images using Kie.ai models' lacks WHEN to use, prerequisites, or output structure. LLMs cannot distinguish generate_image from generate_video without richer context.
No output schemas documented. Callers cannot know what fields generate_image returns (URL? metadata? task_id?). This forces LLMs to guess and breaks tool chaining.
Parameters lack constraints. 'aspect_ratio' and 'resolution' accept free-form strings with no enum or pattern. 'model' parameter has no guidance on valid values or how to discover them (despite list_models existing).
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | <=2025-11-25 | v2 |
upload_file accepts 'path' parameter with no validation hints. No mention of file size limits (512MB exists in config but not documented in tool), allowed formats, or path traversal protection.
No error handling guidance. Tools do not document what errors are retryable, what user-fixable errors look like, or recovery steps. E.g., what happens if a model is not found or API credits are exhausted?