CLI (beta)

The RunLLM CLI lets developers use RunLLM directly from a local terminal. Once installed, run herald from a repository and start asking questions about your code, services, and local environment.

Install the CLI

To try it out, first install uv:

curl -LsSf https://astral.sh/uv/install.sh | sh

Then install the npm package and start the CLI:

npm install -g @herald-ai/herald
herald

Start a Session

From the repository you want RunLLM to use, run:

herald

This opens an interactive session where you can ask questions and continue with follow-ups.

> why are pods in the checkout namespace restarting?

To exit the session, type /exit.

First-Time Setup

The first time you run herald after installing the CLI, RunLLM walks you through a short setup flow. The terminal will show an authentication link, where you can sign in and generate your unique API key. Paste that key back into the CLI when prompted.

During setup, RunLLM inspects what is available on your machine and does its best to connect to useful local tools. For example, if it sees Kubernetes, Google Cloud, Grafana, Prometheus, or similar infrastructure signals in your repo or environment, it can offer to connect to the matching local tools.

These connections use your existing local credentials and configuration. RunLLM does not ask you to manually recreate everything you already have set up on your laptop.

Supported Tools

The CLI can connect to the following local tools during setup or from /config.

Supported today

Local git repositories
Kubernetes kubectl/kubeconfig
Google Cloud gcloud
AWS CLI
Grafana
ClickHouse
MotherDuck
Datadog

Coming soon

GitHub CLI
Azure CLI
Splunk
Elastic
NewRelic
Honeycomb

Available slash commands:

/help shows keyboard shortcuts and available commands.
/history shows questions asked in the current session.
/calls lets you browse tool calls and their full output.
/tools shows configured tool status.
/share exports the session to a Markdown file.
/config lets you edit your API key, repo path, and tools.
/clear clears the conversation.
/exit exits Herald.

Security FAQ

Does RunLLM need my credentials?

The CLI can use local files and tools that are already configured on your machine, such as your kubeconfig, kubectl, and gcloud. You do not need to give RunLLM API keys, cloud credentials, tokens, or service account keys, and those credentials do not leave your local machine.

Can I bring my own LLM key?

Bring your own LLM key support is coming soon. It will be handled as a model configuration option, separate from local tool connections like kubectl, gcloud, and git.

Does RunLLM run locally?

The CLI runs on your local machine and uses your local repository, shell environment, and configured developer tools to gather context. Your existing credentials stay managed by those local tools.

If you run into any issues, please reach out to us at support@runllm.com.