Kansov's MCP Server: Use Claude and Cursor to Manage Your Product Backlog

What if you could type "list my top 5 unaddressed insights from this quarter" directly into Claude Desktop and get a real answer from your actual product data?

Kansov's MCP (Model Context Protocol) server makes this possible. MCP is an open standard developed by Anthropic that allows AI assistants to securely access external tools and data sources. When you connect Kansov to Claude Desktop or Cursor, your AI assistant can read and write your Kansov workspace data — Insights, Ideas, To-Dos, Features, and Knowledge Hub entries — using natural language.

What you can do with the Kansov MCP server

The Kansov MCP server exposes the following tools to your AI assistant:

• list-insights — List recent Insights, filter by product area, source, or date range
• create-insight — Create a new Insight with title, body, source, priority, and customer link
• list-ideas — List Ideas with optional status and product area filters
• create-idea — Create a new Idea with title, description, and scores
• list-todos — List your personal To-Dos, filtered by status or due date
• create-todo — Create a new To-Do with title, due date, and type
• list-features — List Features with optional status filter
• search-knowledge — Semantic search across Knowledge Hub entries
• cross-module-search — Search simultaneously across all modules

What this unlocks for AI-first PMs

The MCP server isn't a parlor trick. It changes the nature of how you interact with your PM system when you're already working in an AI tool.

Morning briefing from Claude

Open Claude Desktop and ask: "What new Insights came in since Monday? Summarize the top themes." Claude calls the Kansov MCP, retrieves your recent insights, and gives you a synthesized briefing — without opening Kansov at all.

Quick capture from anywhere

You're in Cursor writing a feature spec. You realize you need to log a customer signal. Type in Cursor's chat: "Create an insight titled 'Enterprise customer complained about slow report export' with priority High and source Support Ticket." Done — the insight is in Kansov without switching apps.

Cross-referencing during writing

While drafting a PRD in Cursor, ask: "What Insights do we have about the export feature?" Cursor calls the Kansov MCP, retrieves relevant insights, and surfaces them inline in your writing session. No tab switching.

Security and access control

The Kansov MCP server is authenticated and workspace-scoped. You generate an MCP API token from Settings → Integrations → MCP Server — this token is tied to your user account and inherits your RBAC permissions. Your AI assistant can only see and modify data your user account has access to. Tokens can be revoked at any time.

MCP access is disabled by default. Workspace admins control whether MCP is enabled for the workspace.

Setting up the connection

The setup takes about 3 minutes:

1. In Kansov: Settings → Integrations → MCP Server → Generate Token. Copy the token and your MCP URL (https://[your-domain]/mcp).
2. In Claude Desktop: Settings → Developer → Edit Config → add the Kansov server with your URL and token as a bearer header.
3. Restart Claude Desktop and test: ask "List my 5 most recent Insights in Kansov".

For Cursor, the same approach works via Cursor's MCP settings panel. Full setup instructions are in the Integrations documentation.