Skip to main content

EU AI Act Deployer Compliance SDK — wrap Claude/GPT, log everything, generate Art. 26 reports

Project description

auditai

EU AI Act Deployer Compliance SDK — wrap Claude, GPT, Ollama or any OpenAI-compatible LLM and generate Article 26 reports in minutes.

PyPI License: MIT

Install

pip install auditai-sdk

Quickstart

from auditai import wrap_client
import anthropic

client = wrap_client(anthropic.Anthropic(), project="my-app")

# Your code stays identical — every call is now logged and risk-classified
response = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)

Supported providers

Works with any OpenAI-compatible API — including local LLMs:

from openai import OpenAI
from auditai import wrap_client

# OpenAI
client = wrap_client(OpenAI(), project="my-app")

# Ollama (local)
client = wrap_client(
    OpenAI(base_url="http://localhost:11434/v1", api_key="ollama"),
    project="my-app"
)

# LM Studio (local)
client = wrap_client(
    OpenAI(base_url="http://localhost:1234/v1", api_key="lm-studio"),
    project="my-app"
)

# vLLM, llama.cpp, Azure OpenAI — same pattern

CLI

# Risk classification wizard (9 questions → EU AI Act category)
auditai classify

# View call stats
auditai stats --project my-app

# Generate Article 26 Deployer Report (PDF)
auditai report --project my-app --company "Acme SL" --email "cto@acme.com"

# Launch Streamlit dashboard
auditai dashboard --project my-app

What gets logged

Every AI call is recorded in a JSONL audit trail:

{
  "call_id": "uuid",
  "timestamp": "2026-05-07T20:00:00Z",
  "provider": "anthropic",
  "model": "claude-sonnet-4-6",
  "input_tokens": 312,
  "output_tokens": 87,
  "input_hash": "sha256...",
  "output_preview": "first 100 chars...",
  "risk_category": "limited",
  "hitl_required": false
}

Generate compliance report

from auditai import generate_report

report_path = generate_report(
    project="my-app",
    company_name="Acme SL",
    contact_email="compliance@acme.com",
    extra_info={
        "system_description": "Customer support chatbot",
        "use_case": "Automated responses to user queries",
    },
)
# → EU_AI_Act_Report_my-app_2026-05-07.pdf

The report covers Art. 26 obligations: risk classification, technical evidence, HITL events, and deployer declaration.

Links

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

auditai_sdk-0.1.5.tar.gz (23.7 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

auditai_sdk-0.1.5-py3-none-any.whl (24.5 kB view details)

Uploaded Python 3

File details

Details for the file auditai_sdk-0.1.5.tar.gz.

File metadata

  • Download URL: auditai_sdk-0.1.5.tar.gz
  • Upload date:
  • Size: 23.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.4

File hashes

Hashes for auditai_sdk-0.1.5.tar.gz
Algorithm Hash digest
SHA256 380091324f9ce8fe41ddfca37bc2fc988a90b7a1e81a4bb94f3846807fdf9f86
MD5 d77c3a08076146cd4606c633fab9e088
BLAKE2b-256 c2d99844e894c962523a70a31075c889f0c9e7682370ed08772df4932cd0d1bc

See more details on using hashes here.

File details

Details for the file auditai_sdk-0.1.5-py3-none-any.whl.

File metadata

  • Download URL: auditai_sdk-0.1.5-py3-none-any.whl
  • Upload date:
  • Size: 24.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.4

File hashes

Hashes for auditai_sdk-0.1.5-py3-none-any.whl
Algorithm Hash digest
SHA256 fc9b3e48d34d1b3d037dd4c6209c644340d091ee50f0011052999d0a7bac34e3
MD5 e01aad90da9e4a2b2500285d1c295ce7
BLAKE2b-256 c9a9f714446a8163186d96713ec0095284445d377ec22abf9ec299d9825eeb41

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page