KanbanMCP is a VSCode extension that ships an MCP (Model Context Protocol) server so AI assistants like Cline, Roo Code, Codex, Claude Code, Kilo, Kiro, and ZCode.
Highlights
- Many providers: GitLab, Jira, Confluence, Jenkins, Oracle, Kubernetes, Rancher, Kibana, Teams, Figma, and more.
- Auto-config: One click to sync MCP config to Cline/Roo Code/Codex/Claude Code/Kilo/Kiro/ZCode.
- Simple control: Start/stop and configure from the Activity Bar UI.
- Flexible transport: stdio or SSE.
- Vietnamese support: Enable via
kanbanmcp.language. - Proxy isolation: inherited assistant
HTTP_PROXY/HTTPS_PROXYvalues are stripped from the MCP runtime by default; setKANBANMCP_ALLOW_ENV_PROXY=trueonly if KanbanMCP itself must use them.
Quick start
- Install KanbanMCP from VSCode Marketplace.
- Open the KanbanMCP icon on the Activity Bar.
- Click Start MCP Server.
- Click Auto-Config for Cline/Roo Code/Codex/Claude Code/Kilo/Kiro/ZCode to sync MCP config for your assistant.
Figma provider
- KanbanMCP's Figma provider currently uses the Figma REST API with a Personal Access Token.
- Supported workflows include file/node inspection, components/styles/variables reads, asset export, comments, version history, variable writes, and Dev Mode resource link management.
- Variable write tools require
file_variables:readandfile_variables:write. - Dev resource tools require
file_dev_resources:readandfile_dev_resources:write. - Some Figma variable APIs also require edit access plus a compatible Figma plan or seat.
- The current provider does not create or edit Figma canvas content such as frames, layers, or components on the design surface.