Skip to main content

​
Overview

OpenReward integrates seamlessly with GitHub to provide continuous deployment for your environments. When you push code to a connected repository, OpenReward automatically:
  1. Builds your container image
  2. Deploys the new version
  3. Makes your environment available
Prerequisite: You must create an environment before connecting a GitHub repository. See Your First Environment for setup instructions.

​
Connecting a Repository

​
Via OpenReward Web UI

  1. Navigate to Environment Page
    • Go to https://openreward.ai/{username}/{environment-name}
  2. Connect GitHub Repository
    • Click “Connect GitHub” button
    • You will need to login to github, and grant read access to the desired repository
    • Select the repository from dropdown
  3. Configure Build Settings
    • Resource Allocation: CPU and memory for environment
    • Scaling settings: Minimum and maximum number of instances, and the number of sessions per instance
  4. Deploy
    • Click “Deploy” button
    • First build starts immediately

​
Deployment Flow

​
Automatic Deployment on Push

Once connected, deployments happen automatically: 
1. Developer pushes to GitHub
   ↓
2. GitHub webhook notifies OpenReward
   ↓
3. OpenReward validates the push event
   ↓
4. OpenReward builds container image
   ↓
5. New version deployed automatically
   ↓
6. Environment ready with new code

​
Monitoring Deployments

​
Deployment Status

Via Web UI:
  • Navigate to environment page
  • Click “Deployments” tab
  • See list of recent deployments with status:
    • ⏳ Building: Build in progress
    • ✅ Deployed: Successfully deployed
    • ❌ Failed: Build or deployment failed
Deployment Information:
  • Git commit SHA and message
  • Triggered by (push or manual)
  • Build duration
  • Deployment timestamp

​
Build Logs

View in OpenReward:
  1. Go to “Deployments” tab
  2. Click on deployment
  3. View build logs inline

​
Environment Logs

View in OpenReward:
  1. Go to environment page
  2. Click “Logs” tab
  3. See real-time logs from your environment server

​
Resource Configuration

CPU and Memory: Configure via Web UI:
  • Choose appropriate size for your workload
  • Options like “1:2” (1 CPU, 2GB RAM)
Scaling Configuration:
  • Minimum instances: Scales to zero when idle (saves cost)
  • Maximum instances: Limit concurrent environment servers
  • Sessions per instance: How many agents per server

​
Branch and Tag Filtering

Deploy from Specific Branch:
  • Set in repository connection settings
  • Only pushes to this branch trigger deployments
  • Default: main
Deploy from Tags: 
# Tag a commit
git tag v1.0.0
git push origin v1.0.0

# Triggers deployment if configured

​
Best Practices

Reduce image size and build time:
# Build stage
FROM python:3.11 AS builder
COPY requirements.txt .
RUN pip install --user -r requirements.txt

# Runtime stage
FROM python:3.11-slim
COPY --from=builder /root/.local /root/.local
COPY server.py .
ENV PATH=/root/.local/bin:$PATH
CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "8080"]
Catch issues early:
docker build -t test-image .
docker run -p 8080:8080 test-image
curl http://localhost:8080/health
Exclude unnecessary files from build context:
# .dockerignore
.git
.env
*.pyc
__pycache__
tests/
docs/
*.md
Faster builds, smaller images
  • Use slim base images (python:3.11-slim vs python:3.11)
  • Remove unnecessary dependencies
  • Use multi-stage builds
  • Clean up after installations
RUN pip install --no-cache-dir -r requirements.txt
Commit messages appear in deployment history:
# Good
git commit -m "Add caching to reduce response time"

# Bad
git commit -m "fix stuff"

​
Next Steps

Your First Environment

Complete tutorial for deploying your first environment

Local Development

Test environments locally before deploying

Environments

Understand environment architecture