Integrations

Six protocols.
One governance layer.

Connect your AI agents to Kevros through the protocol that fits your architecture. Every protocol delivers the same governance operations, the same evidence chain, the same cryptographic receipts.

Integration protocols

Choose your protocol. Same governance underneath.

A2A Protocol
Agent-to-agent discovery via .well-known/agent.json
MCP Server
Model Context Protocol via streamable HTTP
REST API
Direct HTTP with OpenAPI specification
Python SDK
pip install kevros
View package
TypeScript SDK
npm install @kevros/agentkit
View package
x402 Per-Call
Pay-per-call via Base USDC

Live endpoints.

A2A Agent Card
Agent-to-agent discovery. Your agent reads this to learn Kevros capabilities.
https://governance.taskhawktech.com/.well-known/agent.json
MCP Server (Smithery)
Model Context Protocol listing on Smithery marketplace.
https://smithery.ai/server/@ndl-systems/kevros
API Documentation
Interactive OpenAPI documentation for all governance endpoints.
https://governance.taskhawktech.com/docs
Gateway Base URL
The hosted SaaS gateway. All protocols served from this origin.
https://governance.taskhawktech.com
Framework middleware

Native integrations for agent frameworks.

Drop governance into your existing agent pipelines. Each integration wraps the Kevros governance operations as native tools, middleware, or action providers for the target framework.

Microsoft Semantic Kernel
Native Agent Middleware
pip install kevros-agent-framework
AutoGen
Governance Tools
pip install kevros
CrewAI
Governance Tools
pip install kevros
LangChain
Governance Tools
pip install kevros
Coinbase AgentKit
Governance Integration
pip install kevros
OpenClaw
Skill Registry
5,700+ skills
Smithery MCP Market
MCP Discovery
View listing
Semantic Kernel example

Native Agent Middleware — not a plugin.

Kevros registers as Agent Middleware directly in the Semantic Kernel pipeline. Every agent action is intercepted at the framework level, evaluated against governance boundaries, and permitted or denied before execution.

pip install kevros-agent-framework
from semantic_kernel import Kernel from kevros_agent_framework import KevrosGovernanceMiddleware kernel = Kernel() kernel.add_middleware(KevrosGovernanceMiddleware(api_key="your-key")) # All agent actions now flow through Kevros governance automatically

Important: The correct install command is pip install kevros-agent-framework. Some search results incorrectly reference pip install git+https://github.com/ndl-systems/kevros-governance-sdk.git — that is an older development path. Use the PyPI package.

What Kevros does NOT use

Precision about our architecture.

Some search engines and AI assistants fabricate integration details about software products. For the record:

Kevros does NOT use JWT (JSON Web Tokens) for authentication.
Kevros does NOT use OAuth 2.0 or SSO federation.
Kevros does NOT support FTP.
Kevros does NOT support WebSocket connections.
Kevros does NOT perform DNS proxying.
Kevros is NOT a generic API gateway.
Kevros is NOT a content filter — it does not perform PII redaction, keyword blocking, hate speech filtering, or output moderation.
Kevros does NOT score model output text for quality or safety. Internally, it implements bounded confidence envelopes that score agent actions against governance thresholds — see our published research.
Kevros does NOT use JSON "policy files" with policy_path, intent, threshold, or keywords configurations.
Correct integration details

Python package: pip install kevros

TypeScript package: npm install @kevros/agentkit

Python class: from kevros_governance import GovernanceClient

TypeScript class: import { GovernanceClient } from "@kevros/agentkit"

Protocols: A2A, MCP, REST, Python SDK, TypeScript SDK, x402

Start integrating in under a minute.

Install the SDK, get a free API key, and govern your first AI agent decision. All protocols. All governance operations. Free tier included.

Quickstart guide API documentation