Optio

Optio is an open-source, MIT-licensed workflow orchestration system for AI coding agents. It takes a software development task — typically a ticket from GitHub, GitLab, Linear, Jira, or Notion — and drives it through a complete lifecycle from intake to a merged pull request. The system runs agents in isolated Kubernetes pods, one per repository, using git worktrees for concurrent task isolation. Optio is built for teams that want to scale AI-assisted development without the overhead of manually monitoring agents. Optio supports three modes: tasks that drive tickets to merged PRs, standalone agent workflows triggered on schedule or via webhook, and connections that give agents access to external services (Notion, Slack, Linear, GitHub, PostgreSQL, Sentry) at runtime. Its key differentiator is the autonomous feedback loop: when CI fails, the agent resumes with the failure context; when a reviewer requests changes, the agent picks up the comments and pushes a fix. It repeats until the PR is merged and the issue is closed. The seven-stage pipeline (Intake, Queued, Provisioning, Running, PR Opened, CI & Review, Merged) is fully monitored and self-healing. Tasks auto-resume on failures, auto-merge when CI passes and review is approved, and auto-close linked issues. The real-time dashboard provides live log streaming, pipeline visualization, cost analytics, and cluster health monitoring. Optio is self-hosted and designed for production. It uses Fastify API, Next.js dashboard, BullMQ workers, and Drizzle on Postgres, and ships with a Helm chart for Kubernetes deployment. Setup requires Docker Desktop with Kubernetes enabled, or a full Kubernetes cluster. Compared to managed agent platforms like GitHub Copilot Workspace or Devin, Optio gives you full control over agent configuration, infrastructure, and data — at the cost of higher operational complexity.