Axiomatic
Getting Started

Installation

Install the CLI

curl -fsSL https://axiomatic.sh/api/install | sh

Or install via npm:

npm install -g @exalto/axiomatic

# Or run directly without installing
npx @exalto/axiomatic

Verify the installation:

axm --version

Configure an LLM Provider

Axiomatic needs access to an LLM to analyze your code. You have two options:

Platform (free credits, no API key needed)

Sign in with GitHub to get $5/month in free LLM credits:

axm login

The CLI automatically routes requests through the Axiomatic LLM proxy when you're logged in. You also get access to curated best-practice packs, a dashboard for tracking results, and shared agent memory that cuts costs 60-80%. Check your balance with axm credits. See Billing for plan details and top-ups.

Local (bring your own API key)

Set your provider API key as an environment variable:

# Anthropic (default, recommended)
export ANTHROPIC_API_KEY=sk-ant-...

# Or OpenAI
export OPENAI_API_KEY=sk-...

You can also reference environment variables in your axiomatic.yml config:

# axiomatic.yml
api_key: ${ANTHROPIC_API_KEY}

The CLI works fully standalone with a local API key — no platform account needed. You can connect to the platform later with axm login to get dashboard features and team sharing alongside your own key.

Validate Your Setup

Run the built-in diagnostics to verify everything is configured correctly:

axm doctor

This checks your axiomatic.yml config, axiomatic/ directory, API key or platform credentials, and test files.

You can also lint your test files without running them:

axm validate

Cost Expectations

ModelCost per testBest for
Claude Haiku$0.01-0.05Most tests, fast iteration
Claude Sonnet$0.05-0.20Important tests needing higher accuracy
Claude Opus$0.20+Critical security audits

Caching means you only pay when the relevant code actually changes. A typical project with 10 tests costs a few cents per CI run after the initial analysis.

Next Steps

  • Quickstart — end-to-end walkthrough from install to CI
  • First Test — write and run your first test

On this page