Getting Started with Claude Code on Smarter

Goal

Get your first Claude Code response through Smarter — from both the terminal and the browser — and integrate it into your daily development workflow.

Prerequisites

  • An active Smarter account. If you don’t have one, contact your administrator.

  • Comfort working in a terminal and with command-line tools.

  • A code editor (VS Code recommended).

Note

Your administrator has already configured Anthropic as an LLM provider. You do not need your own Anthropic API key — Smarter manages provider credentials centrally.

Setup

Step 1: Install the Smarter CLI

Follow the installation instructions in Smarter CLI to download the binary for your operating system and add it to your PATH.

Confirm the installation:

smarter version

Step 2: Create a Smarter API Key

Your Smarter API key is your identity on the platform. You need it before you can configure the CLI.

  1. Log in to the Smarter web console.

  2. Click your profile icon (top-right) and select API Keys.

  3. Click Create API Key, give it a name (e.g. my-cli-key), and copy the value.

Set it as an environment variable so you don’t have to type it repeatedly:

export SMARTER_API_KEY=your-api-key-here

Step 3: Configure the CLI

smarter configure

When prompted, provide the target environment (e.g. prod, alpha) and your API key from Step 2.

Concept Overview

Your Smarter API key authenticates every request you make. When you send a prompt, Smarter routes it to the correct provider backend — in this case, Anthropic — handles authentication on your behalf, and returns the response. You never touch Anthropic credentials directly.

Three concepts you need to understand:

  • Provider — a configured LLM backend (Anthropic, OpenAI, etc.). Your administrator manages these.

  • LLMClient — a named resource that bundles a provider, a model, a system prompt, and optional data plugins. This is what you interact with.

  • Manifest — the YAML file that defines any Smarter resource. Every llm_client, provider, and plugin is described this way.

You interact with llm_clients, not providers directly. Your administrator has already created an llm_client backed by Claude Code — you just need its name.

Step-by-Step: Using Claude Code

Step 4: Find Your LLMClient

smarter get llm_clients

This lists every llm_client assigned to your account. Note the name of an llm_client configured with the Anthropic provider.

Step 5: Chat from the Terminal

smarter chat <llm_client-name>

This opens an interactive session with streaming responses. Type your prompt and press Enter.

Useful commands while exploring:

# Confirm the Anthropic provider is active
smarter describe provider anthropic

# Inspect an llm_client's full configuration
smarter describe llm_client <llm_client-name>

Step 6: Chat from the Browser

  1. Navigate to Workbench in the left sidebar of the Smarter web console.

  2. Select your Claude Code llm_client.

  3. Type a prompt and press Enter.

The Workbench lets you tune system prompts, temperature, and max tokens — useful for testing before you automate or deploy.

Step 7: Integrate into Your Workflow

Terminal — pipe code directly to the CLI:

cat my_script.py | smarter chat <llm_client-name> --prompt "Review this code for bugs"

REST API — call Smarter from scripts or CI/CD pipelines. See the Swagger API Documentation for endpoint details.

VS Code — install the Smarter YAML extension for manifest authoring with autocomplete and validation.

Proof of Concept

Run this sequence end to end:

smarter status                    # platform is reachable
smarter get providers             # anthropic is listed and active
smarter chat <llm_client-name>

In the chat session, type:

Explain the difference between a Python list and a tuple in two sentences.

Expected response (approximate):

A Python list is a mutable, ordered collection that can be changed after
creation, while a tuple is immutable and cannot be modified once defined.
Tuples are generally faster and used for fixed data, whereas lists are
preferred when the collection needs to change.

A clear, concise response within a few seconds means your setup is complete.

Troubleshooting

“Not authenticated”

Run smarter configure and re-enter your API key. You can also pass it inline with --api_key.

“No llm_clients found”

Your administrator has not yet assigned an llm_client to your account. Ask them to grant you access to an llm_client configured with the Anthropic provider.

Slow or timed-out responses

Run smarter status. If the platform is healthy, the delay is on Anthropic’s side — wait a minute and retry.

“Provider not available”

The Anthropic provider failed verification. Run smarter describe provider anthropic and check the status section, then contact your administrator.

Unexpected model behavior

Run smarter describe llm_client <llm_client-name> to confirm which model the llm_client is using. Different Claude models have different capabilities and context window sizes.

See also