Figmagent

Analyze a Figma MCP test session transcript. Reads raw session data (JSON or HTML) and produces a structured analysis document with metrics, efficiency issues, error patterns, and prioritized improvements. Updates the cross-session improvement tracker. Use after completing a Figma session or when reviewing past sessions. Accepts an optional file path argument; if omitted, analyzes the most recent transcript.

Figma design knowledge for AI agents operating through MCP tools. Use whenever working in Figma via MCP — creating layouts, building components, managing design tokens/variables, organizing files, or manipulating any design element. Covers component architecture (variants, properties, slots, atomic design), auto layout (direction, spacing, sizing, wrap, absolute positioning), variables and styles (token architecture, modes, scoping, aliasing), collaboration features (comments, annotations, sections, Dev Mode), and critical gotchas (font loading, immutable properties, coordinate systems, instance limitations). Also trigger when the user asks to build a design system, create responsive layouts, set up design tokens, organize a Figma file, or work with component libraries. If the task involves Figma in any way, use this skill.

Build Figma plugins with full Plugin API access. Use when creating plugins for Figma, FigJam, Slides, or Dev Mode that manipulate designs, generate content, export assets, manage styles/variables, work with components, integrate external services, or monetize with payments. Covers manifest configuration, sandbox architecture, UI development with React, node manipulation, auto layout, text/font handling, image processing, variables (design tokens), annotations, codegen, payments, data storage, and publishing. Also trigger for Figma Slides plugin development, Dev Mode codegen plugins, or any work involving the Figma Plugin API.

Orchestrator guide for delegating Figma MCP phases to specialized sub-agents. Use when a Figma task is large enough to risk context overflow — component sets with 8+ variants, unknown tree depth, or sessions expected to exceed 100 tool calls. Supports both serial discovery and parallel build/style phases.