Features
What Easy Claude Code does and where your work actually runs. Read this before connecting a repository or entering an API key — it tells you which of your systems see your code.
Cloud execution
Describe a task and it runs on GitHub Actions runners against the repositories you connect. The agent can read, write and execute code there, then commit or open a pull request. Your Anthropic key is stored and passed to the run as a repository secret.
Your code and instructions are sent to the Claude API during a run, and the workspace is a GitHub runner — not your laptop.
Session persistence
Every session is tracked: decisions made, files modified, tasks completed, and what is next. Close the lid and come back to exactly where you left off — the state lives with the project, not with your terminal.
CLAUDE.md auto-generation
Session state is compiled into CLAUDE.md files, the project context Claude Code reads on startup. Your agent knows what happened last session instead of rediscovering it.
Multi-project dashboard
Track several codebases from one interface. See which projects have active sessions, pending tasks and recent activity, and dispatch work without opening a shell.
Optional local worker
If you would rather keep execution on your own machine, the older local mode still exists: pair a worker running locally with this dashboard and it executes there instead, with full access to your local toolchain. It is optional, it is not the default, and the cloud path does not require it.
Local mode is the fallback path. Anything described on the landing page or in the privacy policy about GitHub Actions applies to the cloud path — read them together.