Deploy your platform on Render

Deploy the platform from your own GitHub repository using the template's Render Blueprint. Start with one always-running API instance and managed PostgreSQL.

You'll need a Render account connected to GitHub, a Render API key, Python 3, OpenSSL, and the model credentials your agents use. These steps create billed resources.

Save your platform in your repository

Create an empty GitHub repository. From the template checkout, point origin at your repository and keep the template as upstream:

git remote rename origin upstream
git remote add origin https://github.com/YOUR-ORG/YOUR-REPO.git

Review git status and commit the agent and configuration changes you want to deploy. Keep credentials and local data out of the commit, then push your branch:

git push -u origin HEAD

Use that branch when creating the Blueprint. The branch from the local setup is my-agent-platform.

Prepare production settings

Create a separate environment file if you do not already have one:

[ -f .env.production ] || cp example.env .env.production

Set OPENAI_API_KEY, RENDER_API_KEY, and any additional model credentials you use. Remove unused placeholder credentials. Keep RUNTIME_ENV=prd or leave it unset. Leave database connection settings to the Blueprint so they point to Render PostgreSQL.

The scripts prefer .env.production when it exists. For an existing deployment, set RENDER_SERVICE_ID and RENDER_POSTGRES_ID to the intended resources. Nonempty exported RENDER_* settings override values in the env file.

Create the Blueprint

In the Render dashboard:

1. Choose New → Blueprint and connect your repository.
2. Select the branch you pushed and use the root render.yaml.
3. Review the API service and PostgreSQL database, then deploy the Blueprint.
4. Supply OPENAI_API_KEY when prompted.

The template sets one API instance and attaches no API disk. Keep the service on an always-running plan so scheduling and queue workers remain available. See Render's Blueprint documentation for the provider workflow.

Production startup requires a JWT verification key. Finish authentication in the next step before treating this deployment as ready.

Configure authentication

From your local checkout, run:

./scripts/render/up.sh

The script locates the service, records its public URL, prepares the MCP connection secret, and asks for the JWT verification key.

1. Open os.agno.com, choose Connect OS, and select Live.
2. Enter the Render URL and enable token-based authorization.
3. Copy the full public verification key into the script's prompt.

The script saves the connection settings to .env.production, uploads application settings, and triggers a deploy. If you skipped the prompt, save the quoted JWT_VERIFICATION_KEY in .env.production and apply it with:

./scripts/render/env-sync.sh

Verify the deployment

Wait for the updated deployment and inspect its startup logs in Render. Then check:

Your Render database starts separately from local PostgreSQL. Ingest the selected product documents into the deployed product knowledge base before testing its answers. Code deployment does not copy your local knowledge records or conversations.

Connect through MCP

Use https://YOUR-RENDER-DOMAIN/mcp as the connector URL in a supported AI application. When prompted, enter the MCP_CONNECT_SECRET saved in .env.production.

This is your platform's MCP endpoint. The Agno documentation connector at https://mcp.agno.com is a separate service used while building.

Ask the connected application to run Product Expert with one of your verified questions. Check the resulting run in AgentOS and verify the caller's permissions before sharing the connection with other people.

Update the platform

The default engineers execute inside the API container. Files they change there can be lost when it is replaced. Before relying on hosted engineering work, configure AgentCore and S3 checkpoints.

Continue to Evaluate your agent.