MCP server + WebSocket bridge that drives the slide-mcp Figma Slides plugin (generate designed Figma slides from GitHub Copilot / Claude).
Single tool 'create_deck' has a well-structured schema with comprehensive parameter documentation and clear descriptions. The tool name follows verb_noun convention (create_deck). Input schema is complete with nested objects (theme, items, table) and proper type definitions. Descriptions are detailed (194 chars for tool, 50-150 chars per param on average). However, output schema is not documented, the tool description does not specify what the Figma plugin returns after execution. Error handling is minimal; no guidance on recovery paths or error categorization. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite the tool being destructive (WRITE risk). Parameter descriptions are thorough but could benefit from explicit format constraints and examples of valid enum values.
Create a Figma Slides deck from a structured specification. Generates designed, layout-varied slides with support for 40+ layout types, 17 design system presets, and rich content (text, images, charts, tables).
Output schema not documented. Tool description does not specify what the Figma plugin returns after creating slides (success confirmation, slide IDs, errors, etc.). LLMs cannot plan downstream actions without knowing the response structure.
No tool annotations despite destructive operation. 'create_deck' modifies Figma state (WRITE risk) but lacks destructiveHint annotation. Agents cannot distinguish safe read-only tools from state-mutating ones.
Error handling lacks recovery guidance. No documented error cases, retryability classification, or actionable error messages. If plugin disconnects or times out, the LLM receives a generic error with no guidance on next steps.
No confirmation/dry-run pattern for irreversible operation. 'create_deck' with replace=true overwrites the current page. Agents should be able to preview or confirm before executing destructive actions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
Enum descriptions lack clarity on visual/functional differences. The 17 theme presets are listed but not explained (e.g., what distinguishes 'sky' from 'ocean'? When should an agent choose 'paper' over 'mono'?). LLMs will guess or default to 'light'.