OpenShip Agents is for teams that already run apps on OpenShip, the open-source platform that builds, deploys, and routes software on your own servers or on OpenShip Cloud. The 0.9.7 release lets you hand a failed deployment or a confirmed service incident to an agent that investigates and can act inside access you grant, including when nobody is watching the dashboard. Incident follow-up no longer has to wait for a person to open a client and start the session. You set the job and the ceiling. The agent can investigate and notify, wait for approval, or, only if you allow it, recover an app and open a code fix on a new branch as a draft pull request, with the patch reviewed before approval. Every run keeps a chat so you can inspect actions and take over. Duplicate incidents cannot repeat the same action. Changing access or pausing a flow cancels pending work, and the assistant stays off until you enable it. On Cloud it runs on the customer server you choose, using your existing Claude Code or Codex subscription through MindWire. The same update adds Slack, Discord, Telegram, and Microsoft Teams, so a linked account can list projects, read private logs, and confirm a redeploy or restart from chat without using the agent. OpenShip Agents is a use of MindWire, the open-source Go runtime released alongside it for putting coding agents behind your own interface and tools. OpenShip is Apache 2.0 and comes from Oblien, built by Abdelhamed Mohamed.