Introduction

Welcome to the Mankinds documentation.

Mankinds is an agentic AI governance platform. Autonomous agents evaluate and monitor AI systems across 8 trust dimensions — privacy, security, accuracy, fairness, explainability, accountability, sustainability and systemic risk — continuously, from development to production, grounded in regulatory knowledge.

The platform combines a dual knowledge graph architecture (70+ regulatory frameworks), autonomous evaluation agents, and structured outputs (scorecards, risk sheets, audit packages) to turn compliance from a manual process into an automated, auditable one.

Key Capabilities

• Offline evaluation — Connect an AI system via API or SDK. Agents run structured test scenarios against the system using ground truth datasets.
• Online evaluation — Connect observability tools (Langfuse, LangSmith, Datadog). Agents analyze real production traces to detect drift and degradation.
• 86 evaluation criteria across 8 dimensions, grounded in EU AI Act, GDPR, ISO 42001, NIST AI RMF and 70+ regulatory frameworks.
• 30+ pre-built connectors — databases, observability platforms, document sources, and AI system endpoints.

Resources

• SDK: JavaScript/TypeScript and Python SDK documentation
• Knowledge Glossary: Key terms definitions for AI evaluation

Quick Start

Choose your language:

In a few minutes, evaluate your first AI system.

Get an API key

Create an account on app.mankinds.io and generate your API key from settings.

Install the SDK

pip install mankinds-sdk

Configure and use

from mankinds_sdk import MankindsClient

client = MankindsClient(api_key='mk_your_api_key')

system = client.systems.create({
  "name": "My Chatbot",
  "description": "Customer support chatbot.",
})

connection = client.connections.create(
  system["id"],
  {
      "connector_type": "custom_api",
      "url": "https://api.example.com/chat",
      "method": "POST",
      "body": {"message": "{{input}}"},
      "input_path": "message",
      "output_path": "reply",
  },
  name="Pre-production API",
  is_default=True,
)
dataset_job = client.datasets.generate(system["id"], {"scenarioCount": 10})
client.jobs.wait(dataset_job["id"])

run = client.evaluations.run({"systemId": system["id"], "connectionId": connection["id"]})
result = client.evaluations.wait(run["evaluation_id"])
print(f"Score: {result['summary']['overall_score']}")

See the full SDK guide for the detailed API reference.

Support

For any questions: [email protected]