Quickstart

2 minutes

BASH

1pip install aurex-sdk

PYTHON

1from aurex_sdk import AurexAuditor, patch_all
2import openai
3
4# 1. Initialize and patch (no manual logging required)
5auditor = AurexAuditor()
6patch_all()
7
8# 2. Use your LLM library exactly as before
9client = openai.OpenAI()
10response = client.chat.completions.create(
11    model="gpt-4o",
12    messages=[{"role": "user", "content": "Explain quantum entanglement."}]
13)
14
15# 3. View the runtime report
16score = auditor.get_score()
17print(f"Efficiency Score: {score.value}/100")
18# => Efficiency Score: 84/100
19# => Flags: ['cache_opportunity: $12.40/month saving on system prompt']

No data leaves your machine by default. Cloud sync is opt-in. The ledger file at ~/.aurex/ledger.jsonl is human-readable JSON and can be inspected at any time. It contains only numeric metadata - never raw prompt text.

SDK v0.3

Aurex SDK Documentation

Aurex is a runtime layer for AI agents. It sits in-process between your application code and LLM provider APIs, routing each call to the right model, compressing context, and recovering from provider failures - automatically, in under 2ms.

Agent-friendly docs

Start at /llms.txt for task → URL routing. Fetch one page per concern (e.g. /docs/reference/config, /docs/reliability/policy) - avoid loading hub pages for implementation detail.