Cloudflare MCP. Full control over edge functions and data lifecycle.

Look, you don't wanna jump between a bunch of dashboards and CLI commands just to manage your edge stuff. This server puts all the Cloudflare core services right where your AI agent can hit 'em with natural language commands. You get programmatic control over everything: Workers, data stores like KV and D1, routes, secrets, and analytics—all through one interface.

Worker Control & Deployment

It's simple to manage your worker lifecycle. Use list_workers to see a list of every script you have running in your account, including when they were last modified. Need more detail? Run get_worker to fetch the full configuration for an existing Worker script. You can check out historical code snapshots using list_worker_versions, or get specific details on one version with get_worker_version.

To see what's been deployed, run list_deployments, which shows you a history of deployments, including version IDs and current traffic percentages.

When it comes to pushing new code live, you control the rollout with create_deployment. You can roll out a specific Worker version either immediately or by setting a gradual percentage increase. If you need to know what's going on right now, list_deployments shows that history. To check if your script is running correctly in real-time, start a live log stream using create_tail_session; when you’re done debugging, use delete_tail_session to close the session.

If a Worker's function changes or needs removal, use delete_worker to completely wipe out the script and all its resources. You can also manage traffic routing by defining URL patterns with create_worker_route, which directs specific traffic to your Worker script; when you're done with that route, use delete_worker_route to stop serving traffic for it.

Data Storage and Queries (KV/D1)

For data persistence, you've got two main options. First, there are KV namespaces, which store key-value pairs. Use list_kv_namespaces to get a list of all available namespaces by ID and title. To see what keys exist inside one of those, run list_kv_keys. When you're ready for the data, use get_kv_key to retrieve the raw JSON value for any specified key.

For structured data, there are D1 databases—serverless SQLite. Start by running list_d1_databases to list all available databases in your account, getting their IDs and names. You execute standard SQL queries (SELECT, INSERT, UPDATE, DELETE) against a specific database using the query_d1 tool.

Object storage is handled through R2 buckets; use list_r2_buckets to see all available object buckets so you know where your files live. Finally, for zone management, run list_zones to get a list of DNS zones associated with your account. If you change static assets or push new code and need the cache cleared everywhere, use purge_cache on an entire zone.

Security and Observability

Keeping things secure means managing credentials properly. Use list_secrets to get a list of all configured secrets for a Worker, showing you names and types without revealing any values. When your app needs a new API key or password, use create_secret to create it or update an existing one; if the secret is no longer needed, run delete_secret.

To track performance, start with general traffic patterns: get_zone_analytics provides aggregated data on CDN performance and traffic for a whole zone over the last 24 hours. For deep worker insights, get_worker_analytics returns metrics like invocation counts and error trends. You can also check out the full deployment history using list_worker_versions, which gives you an audit trail of every code snapshot deployed for a Worker script.

It's all designed to keep your hands on the controls, letting you manage everything from worker deployments and data reads to traffic routing and performance monitoring.