Model Context Protocol (MCP) server for Flipt – expose Flipt feature-flag APIs as MCP tools so AI assistants can list, create, update, delete and evaluate flags.
https://github.com/flipt-io/mcp-server-fliptStop context-switching between your AI chat and Flipt dashboard. This MCP server puts your entire feature flag workflow directly into Claude, ChatGPT, or any MCP-compatible AI assistant.
You're debugging a production issue at 2 AM. You need to check flag states, toggle a problematic feature, or evaluate which users are seeing the new checkout flow. But instead of asking your AI assistant to help analyze the situation, you're tab-switching to Flipt, breaking your flow, and losing context.
Your AI assistant can help you write complex deployment scripts and debug intricate race conditions, but it's completely blind to your feature flag state.
With the Flipt MCP server running, your AI conversations become feature-flag aware. You can:
Your AI assistant becomes a feature flag operations teammate that never sleeps.
Incident Response:
You: "We're seeing checkout errors. What flags are active in the payment namespace?"
AI: [Lists all payment flags with current states]
You: "Disable the new-payment-processor flag immediately"
AI: [Toggles flag and confirms change]
Feature Rollout Management:
You: "How many users are currently seeing the beta dashboard?"
AI: [Evaluates flag distributions and provides user counts]
You: "Increase the rollout to 25% and exclude the enterprise segment"
AI: [Updates rollout rules and confirms changes]
Development Workflow:
You: "Create a flag for the ML recommendation engine, disabled by default"
AI: [Creates flag with proper naming and initial configuration]
You: "Enable it for our internal testing team"
AI: [Sets up segment rules for internal users]
Install via Smithery for Claude Desktop:
npx -y @smithery/cli install @flipt-io/mcp-server-flipt --client claude
Or run it standalone:
npx -y @flipt-io/mcp-server-flipt
Set your Flipt connection:
export FLIPT_API_URL=https://your-flipt-instance.com
export FLIPT_API_KEY=your-api-key
That's it. Your AI assistant now has full Flipt integration.
Feature flags aren't just configuration—they're operational tools that need real-time management. When your AI assistant can see flag states, evaluate user contexts, and make changes on your behalf, you're not just saving time on routine tasks. You're enabling a new class of intelligent operations where your AI can help correlate flag states with performance metrics, suggest rollout strategies, and even automate incident response.
The best part? It feels natural. You're already discussing system behavior with your AI assistant. Now those conversations can actually do something about it.