Keep the code on your own machine
Your runner runs on your own machine, next to your repository, your secrets and your internal services. Evident never runs your code and keeps no copy of your repository — chat traffic is proxied through Evident to reach your machine, and the messages you send are stored so they can be queued while your runner is offline.
What actually crosses the wire
Your chat is the real OpenCode web interface, served through Evident to reach the runner on your machine — so the diffs and file contents you open there travel through Evident on their way to your browser. That is the honest limit of running on your own machine: your code is never executed or copied by Evident, but what you look at in chat passes through it.
What it combines
Local runner
The Evident CLI runs OpenCode on your machine and connects it out to Evident.
Model providers
Connect your own provider. The key is handed to your runner and never exposed beyond it.
Web chat
Shows every message, tool call and file edit as it happens.
A runner can also run in your own AWS account Preview instead of on your machine — booted on demand and suspended when idle, deployed into your account. It is in private preview: talk to us to get set up.
What it will and won't touch
By default your runner accepts no files from Evident at all. If you want it to, you name exactly which directories, and nothing can be written anywhere else on your machine.
Read next
- Local runner — how the CLI starts OpenCode on your machine.
- Model providers — connect your own provider key.
- Quickstart — create a runner and start it — about five minutes.
- AWS runner — Preview: deploy into your own AWS account instead.
Put your coding agent where the work is.
Connect a runner and let it start reviewing your pull requests — in minutes. Free to start.