Loading ToolBench
MCP server for generating 3D mesh assets with textures from text prompts or images
Single tool 'execute_skill_action' has a reasonable name starting with a verb, but critical gaps undermine quality. Description is 87 chars (within 10-1024 range but minimal). Input schema is present with 3 parameters, all with type and description, but output schema is completely undocumented, LLMs cannot plan downstream calls. No error handling guidance, no recovery paths, no validation rules in descriptions. The tool performs a WRITE operation (generates 3D assets) but lacks idempotent hints, confirmation patterns, or destructive warnings. Parameter descriptions lack format constraints (e.g., what image URIs are valid? what polygon ranges?). No security considerations despite accepting external URIs.
Execute the text-to-3D mesh texture asset generation skill
Output schema completely undocumented. LLMs cannot infer what fields are returned or plan downstream tool calls. Baseline: 100% of A+ tools document return types.
Tool description is 87 chars, minimal context. Does not explain WHEN to use it vs alternatives, WHAT it returns, or prerequisites. Baseline: avg 194 chars for production tools.
Parameter descriptions lack format constraints. 'prompt_or_image_uri' does not specify valid URI schemes, image formats, or prompt length limits. 'polygon_count_target' lacks min/max bounds (baseline: numeric params should specify range).
No error handling guidance. Tool performs WRITE (generates assets) but lacks recovery paths, retryability classification, or actionable error messages. LLMs cannot self-correct on failure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
No tool annotations (destructiveHint, idempotentHint). Tool modifies state (creates 3D assets) but LLMs cannot determine if it is safe to retry or if it has side effects.