Recipes

Run Claude Code on ATP

Point Claude Code at the Gateway and drive any allowed model through a project key — no code changes, just environment variables.

Create a project API key

In the console, create (or choose) an organization, workspace, and project, then issue an API key from that project. The key inherits the project's allowed models and credit balance. Copy the full secret — it's shown only once. See Managing API keys.

Point Claude Code at the Gateway

Set the base URL (no /v1 — Claude Code appends /v1/messages itself), send the key as the auth token, and clear ANTHROPIC_API_KEY so it doesn't take precedence.

export ANTHROPIC_BASE_URL="https://api.atptoken.ai"
export ANTHROPIC_AUTH_TOKEN="atp-..."
export ANTHROPIC_API_KEY=""
export ANTHROPIC_MODEL="claude-sonnet-4-6"   # any model from GET /v1/models
claude

Prefer a config file? Put the same values under the env block of ~/.claude/settings.json.

Pick an allowed model

ANTHROPIC_MODEL must be a model id returned by GET /v1/models and enabled for the key's project — otherwise the Gateway returns 403. List the ids first if you're unsure.

Verify

Run a prompt in Claude Code, then open the console: the call shows up under Request logs, and the credits it spent appear on the Usage page. See Usage & logs.

Codex too

The same idea works for Codex and other agents that speak the OpenAI or Anthropic wire format — see Coding agents for the Codex config.

Next →Team with budget caps