Render CLI for coding agents

When a coding agent can run shell commands with access to a Render account, it can use the Render CLI to inspect services and deploys. It can also read logs, query Postgres databases, and manage other Render resources. The CLI can return results as JSON or YAML, formats that make the data easier for an agent to read. The agent can find the service connected to a project and review its current configuration without changing it.

The CLI also has a skills command for managing Render agent skills. A skill gives an AI coding tool instructions for working with Render. The official CLI documentation lists tools that the command supports, including Claude Code, Codex, OpenCode, and Cursor. The docs do not say which skills are available or how to install them, so check the current CLI help and documentation before relying on the command.

Install and authenticate

On macOS, install the CLI with Homebrew:

brew install render-oss/render/render
render --version

The Render docs also list WinGet, curl, direct downloads, and building from Go as installation options. Follow the official installation instructions for those methods.

For interactive use, run:

render login

The CLI opens a Render Dashboard page where you confirm your sign-in in a browser (browser confirmation), then saves the resulting token in its local configuration. For scripts and agent sessions, set the RENDER_API_KEY environment variable to an API key. This variable takes precedence over a token saved by render login, so check which credential the agent's shell receives. Keep the key in your environment or a secret manager; do not paste it into an agent prompt.

List services without changing them

A useful read-only task is to find the Render service connected to a repository before investigating a deployment issue. After authentication, run:

render services list --output json

The command contacts Render and returns service details as JSON. The agent can compare service names and IDs with the repository's deployment configuration, then report which service appears relevant. Scripts and agents can read the JSON more easily than the terminal-formatted output. This command reads account data; it does not create or update a service.

A realistic prompt for this task is:

Use the Render CLI to list my services as JSON. Compare the service names with this repository’s Render configuration, then report the likely matching service and its ID. Do not deploy, update, restart, or delete anything.

The agent reads the repository configuration locally and gets the service list from Render over the network. If the agent cannot identify a match, it should report the ambiguity; matching requires more than a similar name.

Validate before a deploy

The CLI can validate a Render Blueprint, which is a YAML file that describes Render resources. From a directory containing render.yaml, run:

render blueprints validate render.yaml

The validation command checks the Blueprint but does not deploy or update services. When you are ready to deploy, the CLI's deploys create command starts a deploy for the specified service. In non-interactive use, the documented options include --wait to wait for completion, --commit for a Git-backed service, and --image for an image-backed service. Render's deployment documentation describes the deployment behavior.

Know which identity and actions the agent uses

You can use the CLI without interactive prompts in scripts and CI, but the agent's commands use the credentials available to its shell. Since RENDER_API_KEY overrides the token from render login, a shell set up for automation can use a different Render identity from the one you signed in with in the browser.

The CLI can also create or update services, restart them, run one-off jobs, and trigger deployments. Keep those commands out of a read-only investigation prompt. For an agent task that only needs diagnosis, ask it to inspect service details or logs and report its findings without running commands that change Render resources.