mi60dev/visionaire-engine

Overview

Deterministic MCP server for AI coding agents that resolves CSS cascade conflicts. It tells the LLM which rule wins, in which file and on which line, with reasoning. Includes cascade verdicts, blast radius analysis, interaction timelines, and pixel-perfect audits.

Best for

Best for
AI coding agents that need precise CSS debugging context

Use cases

• Debug conflicting CSS rules in component libraries
• Audit visual regressions before merging
• Trace which stylesheet changes affect a given element

How to use

Install

claude mcp add visionaire -- npx -y visionaire-engine

Tools exposed

• page_snapshot
• page_origins
• find_elements
• node_at_point
• pick_element
• inspect_element
• explain_styles
• inspect_ancestors
• get_listeners
• explain_animations
• measure_element
• check_alignment
• pick_color
• record_interaction
• inject_css
• style_diff
• annotated_screenshot
• assert_visual
• visual_diff
• impact_preview

Tested with

Claude Desktop, Claude Code, Cursor

Notes

Deterministic MCP server for AI coding agents that resolves CSS cascade conflicts. It tells the LLM which rule wins, in which file and on which line, with reasoning. Includes cascade verdicts, blast radius analysis, interaction timelines, and pixel-perfect audits.

4 stars on GitHub. Last updated 2026-08-11. Licensed Apache-2.0.

Use cases

• Debug conflicting CSS rules in component libraries
• Audit visual regressions before merging
• Trace which stylesheet changes affect a given element

Pros

• Deterministic output with exact file and line references
• Works with any MCP-compatible coding agent
• Covers blast radius and interaction impacts

Cons

• Early-stage project with minimal community adoption
• Requires MCP setup and configuration
• Pixel-perfect audit scope not documented

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.