Developer Tools & Devops · Fly + Claude
Fly MCP with Claude: Deploy and Monitor Apps Through Natural Language
The short answer
Claude connects to Fly.io through the Model Context Protocol (MCP), an open standard that lets Claude use other software with permissions you control. Once connected, Claude can read app and region status, help deploy containers as micro-VMs, and manage scaling through natural language, while people approve consequential changes before they happen.
What is Fly?
Fly.io turns containers into micro-VMs, small lightweight virtual machines that run in data centres around the world. Developers and operations teams use it to deploy applications close to their users, which reduces delay and keeps services available if one location has trouble. Platform engineers, DevOps teams and software companies of all sizes use it to host web apps, APIs and background services without managing physical servers.
How Claude connects to Fly through MCP
The Model Context Protocol (MCP) is an open standard that works like a universal adapter. It lets Claude securely use other software, reading information and taking actions, with permissions your organisation controls. Fly.io provides a way for Claude to connect using an API key, so the connection is set up deliberately by your team.
Once connected, Claude can typically look up your apps, check the status of micro-VMs, view which regions are running and healthy, and help with deployments and scaling changes. You can ask in plain English, for example to check whether an app has enough capacity in Singapore before a busy period.
Permissions and approvals keep people in charge. We grant least-privilege access, so Claude only sees and changes what it needs. Consequential actions such as deploying to production or scaling down can be set to require a named person's approval, and every action is recorded in an audit log.
What you can automate
Release status checks for engineering leads
Before a release, an engineering lead can ask Claude which apps are running which versions, whether any micro-VMs are unhealthy, and which regions are reporting issues. Claude gathers the status across apps and presents a short summary, saving the time usually spent switching between dashboards and chat threads.
Regional health reports for operations
Operations teams can have Claude summarise region status each morning, highlighting any location where latency or availability looks unusual. The report can be sent to the operations channel, so the team starts the day knowing where attention is needed.
Scaling requests from the business
When a marketing campaign or seasonal sale is expected, a manager can describe the demand in plain language. Claude proposes the scaling changes needed for each app and region, and an engineer approves them before anything changes.
Deployment support for small development teams
Smaller teams without a dedicated platform engineer can use Claude to walk through deployment steps, check configuration against what is running, and confirm that a new version came up correctly. Consequential steps still wait for a human to approve them.
Incident triage for support teams
When customers report slow pages, a support or on-call engineer can ask Claude to check micro-VM health and recent region status for the affected app. This shortens the first investigation and helps the right person be called sooner.
Cost and capacity reviews
Finance and engineering can review how many micro-VMs each app runs and where they sit. Claude can help produce a clear list of over-provisioned apps, giving both teams a shared starting point for cost discussions.
An end-to-end workflow: Campaign launch with capacity check
TriggerA marketing manager logs a planned product launch in the project tracker in Microsoft 365, with the expected traffic and target countries.
Claude reads the launch details from the tracker and identifies the apps involved.
Claude checks current micro-VM counts and region status on Fly.io for those apps.
Claude drafts a scaling plan for the target regions and notes any region showing warning signs.
The engineering lead reviews the plan and approves the changes in the approval queue.
Claude applies the approved scaling changes, confirms the new state, and records the actions in the audit log.
Claude posts a summary to the launch channel in Microsoft 365 for the marketing and support teams.
OutcomeThe launch team has the capacity they need in the right regions, engineers keep control of every change, and everyone has a clear record of what was done and why.
How we set it up
Free automation audit
We begin with a free audit of your current deployment and operations work, identifying the workflows around Fly.io that are worth automating and the people who should approve changes.
Secure connection with an API key
We connect Claude to Fly.io using an API key, issued and stored securely with least-privilege scope, so access matches only what the workflows need.
Workflow design across your systems
We design the end-to-end workflow, linking Fly.io with systems such as Microsoft 365, your CRM, ticketing or ERP where useful, and define which actions need human approval.
Testing and launch with monitoring
We test against real past deployments and incidents, then launch with monitoring dashboards and audit logs so you can see exactly what Claude does.
Team training and handover
We train your engineers, operations staff and managers on how to use the workflows, and hand over documentation so your team can run and adjust them confidently.
Who benefits
- Engineering leads
- Faster status checks across apps and regions, with approvals kept in their hands.
- Operations and site reliability teams
- Daily region health summaries and quicker first-response triage during incidents.
- Product and marketing managers
- Plain-language capacity requests for launches, without waiting for a ticket to be read.
- Finance and IT leaders
- Clearer visibility of how many micro-VMs run where, supporting cost and capacity decisions.
Security and control
- Least-privilege API key access, limited to the apps and actions each workflow needs.
- Named human approval for consequential actions such as production deployments and scaling reductions.
- Full audit logs of every action Claude takes through Fly.io, available for review.
- Testing against real past cases before launch, with a clear rollback plan agreed in advance.
- Regular review of permissions as your apps and teams change.
For teams in the US, UK, India and the Middle East
Organisations in the United States and the United Kingdom can host their Claude workflows in those regions to align with local data expectations. Companies in India can keep processing in India to support obligations under the Digital Personal Data Protection Act, 2023. Organisations in the UAE and Saudi Arabia can choose hosting that suits their local data protection laws. We select the hosting region with you during the audit, so the setup meets your legal and contractual requirements.
Learn more about Claude and MCP
Anthropic's AI model and assistant.
Secure links between Claude and business tools.
How engineers build and change software with Claude.
Hosted agents that run multi-step work.
The open standard Claude uses to connect to tools.
Frequently asked questions
What does the Fly MCP server do?
The Fly MCP server is the bridge that lets an AI assistant such as Claude work with your Fly.io account. It allows the assistant to look up apps, check micro-VM and region status, and help with deployments and scaling, using the permissions you grant. It gives the assistant a structured way to read and act, rather than relying on copied screenshots or pasted commands.
Is connecting Claude to Fly.io secure?
It can be, when set up carefully. We use an API key with least-privilege scope, so Claude only has access to the apps and actions your workflow needs. Consequential changes can require named human approval, and every action is recorded in an audit log. We also review permissions regularly as your environment changes.
What can Claude do with Fly.io once connected?
Typically, Claude can check which apps are running, view micro-VM health and region status, help plan and carry out deployments, and manage scaling through natural language requests. It can also summarise the results for your team and post updates to the tools where people already work. The exact actions are defined in each workflow and limited to what you approve.
How long does setup take?
A focused first workflow often takes a few weeks, from the free audit through testing to launch and training. Setup time depends on how many apps, regions and connected systems are involved, and on how quickly approval rules are agreed. We give you a clear timeline after the audit.
What access do we need to set up the connection?
Generally, you need an API key from your Fly.io account with the scope required for the workflows you choose. Our team recommends issuing a dedicated key for this purpose, limited to the necessary permissions, and rotating it on a regular schedule. We confirm the exact requirements during the audit so access is set up correctly from the start.