原文
Autonomous CI/CD for AI coding agents.
Optio turns coding tasks into merged pull requests — without human babysitting. Submit a task (manually, from a GitHub Issue, or from Linear), and Optio handles the rest: provisions an isolated environment, runs an AI agent, opens a PR, monitors CI, triggers code review, auto-fixes failures, and merges when everything passes.
The feedback loop is what makes it different. When CI fails, the agent is automatically resumed with the failure context. When a reviewer requests changes, the agent picks up the review comments and pushes a fix. When everything passes, the PR is squash-merged and the issue is closed. You describe the work; Optio drives it to completion.
Dashboard — real-time overview of running agents, pod status, costs, and recent activity
Task detail — live-streamed agent output with pipeline progress, PR tracking, and cost breakdown
You create a task Optio runs the agent Optio closes the loop
───────────────── ────────────────────── ──────────────────────
GitHub Issue Provision repo pod CI fails?
Manual task ──→ Create git worktree ──→ → Resume agent with failure context
Linear ticket Run Claude Code / Codex Review requests changes?
Open a PR → Resume agent with feedback
CI passes + approved?
→ Squash-merge + close issue
- Intake — tasks come from the web UI, GitHub Issues (one-click assign), or Linear tickets
- Provisioning — Optio finds or creates a Kubernetes pod for the repo, creates a git worktree for isolation
- Execution — the AI agent (Claude Code or OpenAI Codex) runs with your configured prompt, model, and settings
- PR lifecycle — Optio polls the PR every 30s for CI status, review state, and merge readiness
- Feedback loop — CI failures, merge conflicts, and review feedback automatically resume the agent with context
- Completion — PR is squash-merged, linked issues are closed, costs are recorded
The core differentiator. Optio doesn't just run an agent and walk away — it monitors the PR and drives it to completion:
- Auto-resume on CI failure — agent is re-queued with the names of failed checks
- Auto-resume on merge conflicts — agent is told to rebase and force-push
- Auto-resume on review feedback — reviewer comments are passed as the resume prompt
- Auto-merge — when CI passes and reviews are approved, the PR is squash-merged
- Auto-close issues — linked GitHub Issues are closed with a comment linking the merged PR
One long-lived Kubernetes pod per repository. The pod clones the repo once, then stays alive. Each task gets its own git worktree, so multiple tasks run concurrently without interference.
- Multi-pod scaling — repos can scale beyond a single pod (
maxPodInstancesup to 20) - Persistent volumes — installed tools and caches survive pod restarts
- Idle cleanup — pods are reclaimed after 10 minutes of inactivity (configurable)
- Health monitoring — crashed/OOM-killed pods are auto-detected and restarted
- Worktree lifecycle — automatic cleanup with grace periods for retries
Optio can automatically launch a review agent as a subtask of the original coding task:
- Triggered on CI pass, on PR open, or manually
- Runs with a separate review-specific prompt and model (use a cheaper model for reviews)
- Blocking — the parent task waits for the review to complete before merging
- Configurable per-repo with custom review prompt templates
Each repository can be individually configured:
- Claude model — Opus or Sonnet, with context window (200k or 1M), thinking on/off, effort level
- Container image — auto-detected from repo contents (Node, Python, Go, Rust, Full) or custom Dockerfile
- Prompt templates — Handlebars-style templates with
{{variables}}and{{#if}}conditionals - Concurrency — max concurrent tasks per repo, max pods per repo, max agents per pod
- Extra packages and setup commands — apt packages and shell commands run at pod startup
.optio/setup.sh— repo-level setup script run after clone
- Priority queue — integer priorities with drag-to-reorder
- Bulk operations — retry all failed, cancel all active
- Subtask system — child tasks, sequential pipeline steps, and review subtasks
- Cost tracking — per-task cost in USD, with analytics dashboard (daily trends, per-repo breakdown, top tasks)
- Error classification — failures are categorized (auth, network, timeout, resource, etc.) with human-readable descriptions and suggested remedies
- GitHub Issues — browse issues in the UI, one-click assign to Optio, auto-label and comment back with PR links
- Linear — fetch actionable tickets, sync status, add comments
- Multi-provider OAuth — GitHub, Google, and GitLab authentication for the web UI
- Claude Code + OpenAI Codex — API key or Max Subscription (OAuth token) auth
- Live log streaming — structured agent output streamed via WebSocket
- Pipeline progress — visual stage tracker (queued → setup → running → PR → CI → review → merge → done)
- Event timeline — full audit trail of state transitions
- Cluster view — pod status, resource usage, health events
- Cost analytics — daily cost trends, per-repo and per-type breakdowns, period comparisons
┌──────────────┐ ┌────────────────────┐ ┌──────────────────────────┐
│ Web UI │────→│ API Server │────→│ Kubernetes │
│ Next.js │ │ Fastify │ │ │
│ :3100 │ │ │ │ ┌── Repo Pod A ──────┐ │
│ │←ws──│ Workers: │ │ │ clone + sleep │ │
│ Dashboard │ │ ├─ Task Queue │ │ │ ├─ worktree 1 ⚡ │ │
│ Tasks │ │ ├─ PR Watcher │ │ │ ├─ worktree 2 ⚡ │ │
│ Repos │ │ ├─ Health Mon │ │ │ └─ worktree N ⚡ │ │
│ Cluster │ │ └─ Ticket Sync │ │ └────────────────────┘ │
│ Costs │ │ │ │ ┌── Repo Pod B ──────┐ │
│ Issues │ │ Services: │ │ │ clone + sleep │ │
│ │ │ ├─ Repo Pool │ │ │ └─ worktree 1 ⚡ │ │
│ │ │ ├─ Review Agent │ │ └────────────────────┘ │
│ │ │ └─ Auth/Secrets │ │ │
└──────────────┘ └─────────┬──────────┘ └──────────────────────────┘
│ ⚡ = Claude Code / Codex
┌──────┴──────┐
│ Postgres │ Tasks, logs, events, secrets, repos
│ Redis │ Job queue, pub/sub, live streaming
└─────────────┘
┌─────────────────────────────────────────────────┐
│ INTAKE │
│ │
│ GitHub Issue ───→ ┌──────────┐ │
│ Manual Task ───→ │ QUEUED │ │
│ Ticket Sync ───→ └────┬─────┘ │
└───────────────────────────┼──────────────────────┘
│
┌───────────────────────────┼──────────────────────┐
│ EXECUTION ▼ │
│ │
│ ┌──────────────┐ ┌─────────────────┐ │
│ │ PROVISIONING │───→│ RUNNING │ │
│ │ get/create │ │ agent writes │ │
│ │ repo pod │ │ code in │ │
│ └──────────────┘ │ worktree │ │
│ └───────┬─────────┘ │
└────────────────────────────────┼──────────────────┘
│
┌─────────────┐ │ ┌─────────────────┐
│ FAILED │←──┴──→│ PR OPENED │
│ │ │ │
│ (auto-retry │ │ PR watcher │
│ if stale) │ │ polls every 30s │
└─────────────┘ └────────┬─────────┘
│
┌─────────────────────────────────────────────┼────────┐
│ FEEDBACK LOOP │ │
│ │ │
│ CI fails? ────────→ Resume agent ←──────┤ │
│ to fix build │ │
│ │ │
│ Merge conflicts? ──→ Resume agent ←──────┤ │
│ to rebase │ │
│ │ │
│ Review requests ───→ Resume agent ←──────┤ │
│ changes? with feedback │ │
│ │ │
│ CI passes + ───────→ Auto-merge ───────┘ │
│ review done? & close issue │
│ │
│ ┌─────────────┐ │
│ │ COMPLETED │ │
│ │ PR merged │ │
│ │ Issue closed│ │
│ └─────────────┘ │
└───────────────────────────────────────────────────────┘
- Docker Desktop with Kubernetes enabled (Settings → Kubernetes → Enable)
- Node.js 22+ and pnpm 10+
# Clone and install
git clone https://github.com/jonwiggins/optio.git && cd optio
pnpm install
# Bootstrap infrastructure (Postgres + Redis in K8s, migrations, .env)
./scripts/setup-local.sh
# Build the agent image
docker build -t optio-agent:latest -f Dockerfile.agent .
# Start dev servers
pnpm dev
# API → http://localhost:4000
# Web → http://localhost:3100The setup wizard walks you through configuring GitHub access, agent credentials (API key or Max Subscription), and adding your first repository.
apps/
api/ Fastify API server, BullMQ workers, WebSocket endpoints,
review service, subtask system, OAuth providers
web/ Next.js dashboard with real-time streaming, cost analytics
packages/
shared/ Types, task state machine, prompt templates, error classifier
container-runtime/ Kubernetes pod lifecycle, exec, log streaming
agent-adapters/ Claude Code + Codex prompt/auth adapters
ticket-providers/ GitHub Issues, Linear
images/ Container Dockerfiles: base, node, python, go, rust, full
helm/optio/ Helm chart for production Kubernetes deployment
scripts/ Setup, init, and entrypoint scripts
Optio ships with a Helm chart for production Kubernetes clusters:
helm install optio helm/optio \
--set encryption.key=$(openssl rand -hex 32) \
--set postgresql.enabled=false \
--set externalDatabase.url="postgres://..." \
--set redis.enabled=false \
--set externalRedis.url="redis://..." \
--set ingress.enabled=true \
--set ingress.hosts[0].host=optio.example.comSee the Helm chart values for full configuration options including OAuth providers, resource limits, and agent image settings.
| Layer | Technology |
|---|---|
| Monorepo | Turborepo + pnpm |
| API | Fastify 5, Drizzle ORM, BullMQ |
| Web | Next.js 15, Tailwind CSS 4, Zustand |
| Database | PostgreSQL 16 |
| Queue | Redis 7 + BullMQ |
| Runtime | Kubernetes (Docker Desktop for local dev) |
| Deploy | Helm chart |
| Auth | Multi-provider OAuth (GitHub, Google, GitLab) |
| CI | GitHub Actions (format, typecheck, test, build-web, build-image) |
| Agents | Claude Code, OpenAI Codex |
See CONTRIBUTING.md for development setup, workflow, and conventions.