Coding Agent

Overview

Release, deploy, and track without leaving your coding agent. No dashboard, no commands to memorize, no context switch.

What you can do

Your coding agent knows your Fly registry: every release, every environment, every team decision. Anything you can do in Fly, it can do too.

  • Develop with Agent context. Your agent has every past team decision at hand. As you work, the relevant ones surface. Your own conversations join automatically as Decision Records on the release they shape.
  • Release. Ship a new version with full traceability. If your repo or CI isn’t set up for Fly yet, your agent will handle it for you.
  • Runtime Deploy & Track. Deploy releases to any environment, and see what’s running. “Deploy the login fix to staging.” “Is the login fix live in production?”
  • Distribute. Ship artifacts to customers, partners, or the public. “Distribute the new SDK to our partner channel.” Or have your agent set up your CI to distribute automatically on each release.
  • Upload and download artifacts locally. Move binaries, installers, or one-off files. “Upload my docker image to Fly.” “Get the artifact from yesterday’s release.”
  • Find anything by content. Search releases, deployments, and what’s running across your environments semantically. “Find the release where I fixed the login bug.” “What’s running in production right now?”
  • Manage notifications. Add, edit, or remove watches directly from your coding agent. “Watch for security fixes that reach prod.”
Note

Works with Cursor, VS Code (Copilot), Claude Code, and OpenCode. Requires the Fly App installed with the agent toggle active.

How it works

When the Fly App connects your agent, it adds Fly’s MCP tools (direct actions on Fly) and Fly skills (guided flows like deploying to k8s or configuring CI workflows).

That means you can ask your agent anything you’d do in Fly:

  • “Is the login fix in production?”
  • “Distribute this version to our partners.”
  • “Notify me when David deploys to production.”

Next steps