Loading ToolBench
The server has 4 tools with complete Zod schemas and basic descriptions. All tools follow verb_noun naming (get_*) and have typed parameters. However, descriptions are brief (10-50 chars) and lack guidance on when to use each tool, what to do with the output, or error recovery. Parameter descriptions exist but are minimal. Output schemas are documented as text content only, without field-level detail. No error handling guidance for failures (e.g., Figma API timeouts, missing files). The server is functional but below production quality for agentic use.
Get D-Zero frontend coding guidelines
Get data as a cached JSON file path and content from Figma URL
Get image URL from Figma node ID
Get task step
Descriptions lack guidance on when to use each tool and what to do with results. 'Get data as a cached JSON file path and content from Figma URL' (get_figma_data) is unclear about return format (file path vs inline content?) and when to prefer it over get_figma_image.
Parameter descriptions are minimal and lack format/constraint guidance. Example: 'Task file path (Relative to cwd)' does not explain whether the file must exist, what happens if it doesn't, expected file format (YAML, JSON, MD?), or maximum file size.
No error handling guidance. Tools return text content on success, but what happens on failure? If get_figma_data hits a 429 (rate limit) or 401 (auth), does it return an error message? Can the agent retry? Is the error retryable or user-fixable?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 63 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 45 | - | v1 |
Output schemas are documented as generic 'text' content. Actual field names and types are not exposed. E.g., get_figma_data promises 'content' but is it a string, object, or array? What fields does the cached JSON contain?
No dependency hints or chaining guidance. For example, get_figma_data documentation should note: 'If you need specific nodes as images, call get_figma_image() with the nodeId from this data.' This guides multi-step planning.
Parameter 'techTypes' in get_coding_guidelines is an enum, which is good, but the description repeats the enum values ('html, css, js, media, web-components') rather than explaining what each means or when to select multiple vs. single.