Prompts

Vibma includes 4 built-in MCP prompts that provide best-practice guidance to AI agents. Agents can reference these prompts for context when working with Figma designs.

design_strategy

Best practices for working with Figma designs. Covers:

• Understand before creating — inspect documents, discover existing styles and variables
• Use design tokens — prefer fillStyleName, textStyleName, and variable bindings over hardcoded values
• Auto-layout first — use layoutMode on every container, avoid absolute positioning
• Naming conventions — semantic names, Property=Value pattern for components
• Variable modes — pin frames to specific modes (e.g. Dark)
• Quality checks — run lint_node to catch hardcoded colors, missing text styles, and layout issues

read_design_strategy

Best practices for reading Figma designs. Focuses on using get_selection() and selection-based workflows to understand existing designs.

text_replacement_strategy

A systematic approach for replacing text in Figma designs:

1. Analyze structure — scan text nodes, identify logical groupings (tables, lists, cards)
2. Strategic chunking — divide work by structure, spatial area, semantics, or components
3. Progressive replacement — clone nodes, replace chunk by chunk, export and verify each step
4. Intelligent handling — adapt to container constraints, maintain hierarchy, scale exports appropriately

swap_overrides_instances

Guide for transferring content overrides between component instances:

1. Identify source and target instances
2. Extract overrides with instances(method: "get")
3. Apply matching overrides (text, fills, styles) to target children
4. Verify with get_node_info and export_node_as_image