Cloudflare recently introduced thirteen new Model Context Protocol (MCP) servers—unveiled in June 2025—to help AI-driven tools tap directly into its edge services. Instead of handing over blanket permissions, each AI agent gets just the APIs and data it needs, making automation safer, faster, and far more reliable.
How MCP Servers Work
Think of an MCP server as a miniature, specialized runtime. It exposes only the capabilities an AI model truly requires, so you avoid unnecessary risks and reduce clutter in your automation pipelines. This approach echoes the zero-trust security model, keeping AI interactions as precise and locked-down as possible.
Developer Productivity MCPs
- Documentation Server: Embeds Cloudflare’s Developer docs right inside your AI client—no more switching tabs to look up API specs.
- Workers Bindings Server: Simulates production features like KV storage, R2 buckets, Durable Objects, secrets, and AI models, so you can test Cloudflare Workers before you push any code.
- Container Server: Launches on-demand containers for experimentation or CI/CD tasks, all isolated from your local setup and live environments.
- Browser Rendering Server: Offers headless browsing, screenshot capture, and markdown conversion, perfect for automated UI checks or extracting page content.
Observability & Monitoring MCPs
- Workers Observability Server: Delivers real-time logs, analytics, and error traces so you can pinpoint performance snags in an instant.
- Logpush Server: Monitors your Logpush jobs and flags misconfigurations or delivery hiccups.
- Digital Experience Monitoring Server: Combines real-user and synthetic data to give AI-driven insights on latency, uptime, and error rates.
- DNS Analytics Server: Reports query volumes, response times, and DNS errors to help agents troubleshoot straight away.
Security, Compliance & Intelligence MCPs
- Radar Server: Leverages global internet traffic data to spot outages, scans, and suspicious domains.
- Audit Logs Server: Tracks user actions and configuration changes—essential for compliance audits and forensic deep dives.
- Cloudflare One CASB Server: Acts as a Cloud Access Security Broker, scanning your SaaS applications for risky settings under a zero-trust framework.
AI-Specific Infrastructure MCPs
- AI Gateway Server: Records every prompt and response through Cloudflare’s AI Gateway, measuring token usage, latency, and content to help you balance cost and quality.
- AutoRAG Server: Connects with retrieval-augmented generation systems by indexing and querying your documents, making sure AI agents always have the right context.
Open Source & Integration
All thirteen MCP servers are available on GitHub (github.com/cloudflare/mcp-server-cloudflare) and work with any MCP-compatible client, including Claude.ai. Their open-source nature invites community contributions—imagine tailored MCPs for healthcare, finance, or any other niche.
Why It Matters
For teams building AI-native services, these MCPs are a game-changer. They simplify how AI agents collaborate with cloud infrastructure, cutting down on manual steps and reducing the risk of over-privileged access. As more workflows become autonomous, these focused, auditable interfaces will be vital for balancing innovation with security.
Conclusion: A New Chapter in DevOps and SecOps
These MCP servers aren’t just another set of APIs—they lay the groundwork for next-generation DevOps and SecOps powered by AI. By providing tightly controlled runtimes, Cloudflare is helping organizations scale responsibly, keeping security and compliance at the forefront of every automated flow.