Loading ToolBench
Server for Codigma MCP - A service for converting Figma designs to code
This server has critical definition quality gaps. Tool names lack action verbs (e.g., 'map-figma-node' should be 'convert-figma-node-to-model'). Descriptions are present but minimal (10-60 chars), below the 50-200 char baseline for LLM optimization. Input schemas exist but are severely incomplete: most parameters lack type definitions (e.g., 'absoluteBoundingBox' and 'styles' in map-figma-node are typed as 'object' with no nested schema). Output schemas are entirely undocumented, no tool describes what it returns. Error handling is absent; tools return generic 400/500 responses with no recovery guidance. No parameter validation, no enum constraints, no format specifications. The pipeline tool (figma-to-bundle) combines three concerns (fetch, map, generate) into one, violating single-responsibility. No security considerations visible (Figma token handling is basic). Overall, this reads as a proof-of-concept, not production-grade tooling.
Fetches Figma file data using a Figma URL and personal access token
Complete pipeline that fetches Figma data, maps it to CodigmaModel, and generates an HTML bundle with CSS
Generates a complete HTML bundle with embedded CSS from a CodigmaModel structure
Generates CSS from a CodigmaModel structure
Generates HTML from a CodigmaModel structure
Maps a Figma node to a CodigmaModel structure
Tool names lack action verbs. 'map-figma-node' and 'generate-html' are vague; should be 'convert-figma-node-to-codigma-model' and 'render-codigma-model-as-html'. LLMs cannot infer intent from noun-heavy names.
Input schemas use 'object' type for complex parameters (absoluteBoundingBox, styles, codigmaModel) with no nested property definitions. LLMs cannot validate structure or infer required fields. Schemas must define all nested properties with types.
No output schemas documented. Tools return JSON responses (figmaData, codigmaModel, html, css, bundle) but LLMs have no specification of return structure, field types, or required fields. This breaks downstream tool chaining.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 23 | 2026-07-28+ | v2 |
Descriptions are too short (10-60 chars, below 50-200 baseline). 'Fetches Figma file data using a Figma URL and personal access token' lacks context on when to use this vs the pipeline tool, what data is returned, or error recovery steps.
figma-to-bundle combines fetch, map, and generate into one tool. This violates single-responsibility and prevents agents from composing steps independently. Split into separate tools or document when to use pipeline vs individual steps.