QBIC is a protocol-native AI creation stack powered by MCP for orchestrating tools, agents, and interfaces through language.
QBIC is a next-generation agent orchestration framework that enables intelligent agents to control and coordinate actions across creative and technical tools in real time using the Model Context Protocol (MCP).
It allows developers and AI systems to construct natural language-driven workflows, where tools like Figma, Cursor, IDE environments, and more can be interconnected through a shared protocol layer. Agents communicate via MCP to query context, modify content, and trigger multi-step actions across systems—bridging the gap between intent, interface, and execution.
While QBIC is a broad framework, we currently demonstrate one of its abilities: controlling Figma design workflows through Cursor's AI assistant, showcasing how dialogue-based programming can manipulate layout, text, components, and annotations. This is just one application—QBIC is designed for far more.
In our featured demo:
Cursor connects to QBIC via MCP
Cursor receives user dialogue and issues MCP commands
Figma plugin reacts to tool-level operations such as:
create_rectangle
, create_text
, scan_text_nodes
set_multiple_text_contents
, get_reactions
, create_connections
This scenario illustrates design automation, but similar logic applies to document generation, IDE manipulation, or task automation.
QBIC consists of:
src/qbic_mcp/
) — Core server for executing structured MCP commands and managing tool schemassrc/cursor_mcp_plugin/
as Figma example) — Tool-specific bridges that expose runtime controls and statesrc/socket.ts
) — Event pipe for real-time messaging between agents and connected environments"Generate a 3-section homepage with hero, features, and footer."
"Replace all placeholder copy with production text."
scan_text_nodes
and set_multiple_text_contents
"Convert prototype connections into FigJam-style connectors."
get_reactions
, styles connections, and applies create_connections
"Scan all design annotations and classify them."
get_annotations
with structured metadata taggingcurl -fsSL https://bun.sh/install | bash
bun setup
This registers QBIC in your Cursor's MCP config.
bun socket
bunx qbic-mcp
src/cursor_mcp_plugin/manifest.json
get_document_info
, get_selection
, get_node_info
, read_my_design
create_rectangle
, create_text
, create_frame
set_text_content
, set_multiple_text_contents
get_annotations
, set_annotation
, set_multiple_annotations
annotation_conversion_strategy
set_layout_mode
, set_padding
, set_item_spacing
set_layout_sizing
, resize_node
set_fill_color
, set_stroke_color
, set_corner_radius
clone_node
, delete_multiple_nodes
get_reactions
, create_connections
, set_default_connector
export_node_as_image
design_strategy
, reaction_to_connector_strategy
QBIC is not a plugin. It is a multi-agent, cross-environment automation platform built on standard protocols. It empowers:
Whether it's Figma today or VSCode, Photoshop, or Slack tomorrow — QBIC aims to become the protocol layer of agent-centric creation.
MIT License
No configuration available
Related projects feature coming soon
Will recommend related projects based on sub-categories