Hi Product Hunt. I am Jaymin, and I built Warren to operate coding-agent runs beyond a developer terminal. Coding agents are tools. Warren turns them into infrastructure. Warren becomes useful when an agent run stops being a terminal session and starts being a workload. It may need to continue after your laptop closes, repeat on a schedule, survive failure, or stay within a budget. It may also need to become visible to someone besides the person who started it. Warren owns the operational work around an agent harness: - Workspace. Warren materializes the repository and an isolated place to work. - Run. One lifecycle dispatches the workload locally, in a sibling Docker container, or in a Kubernetes pod. - Limits. Spend and concurrency boundaries hold while the work runs. - Events. Warren persists and streams structured run history. - Intervention. An operator can cancel any live run. Supported runtimes also accept mid-run steering. - Recovery. Warren reconciles failed runtimes and salvages recoverable work. - Delivery. Warren pushes the resulting branch and can open a pull request through its forge integration. Warren runs Warren. It operates runs against its own repository and publishes the history at https://app.warren.run. You can inspect real projects, runs, and live event streams without a login. The run model is harness-independent, but it is not integration-free. Any harness with a Warren runtime adapter can use it. The current distribution ships Pi and Claude Code adapters. Warren is self-hosted, MIT licensed, and pre-1.0 (0.18.0). Start with one operator. A small trusted team can share one deployment trust boundary. Named users, RBAC, and per-user attribution have not shipped. Warren is GitHub-first today, and runtime capabilities differ. I want to hear when an agent run becomes a workload for you. What would stop you from operating it on your own infrastructure?

Warren

Maker

📌

16h ago

Went and looked at the public run history and the useful part is the cost column. On your own warren repo the runs from the last eight hours go from $0.538 to $4.47, same project, same day. That spread is the thing nobody building on top of agents can price around, and publishing it with no login does more for me than the seven bullets above it.

25m ago

Warren

Infrastructure for coding-agent workloads

Infrastructure for coding-agent workloads