Repolode
Submit
yaalalabs

yaalalabs/agent-kernel

A Python platform for deploying and running enterprise AI agents across frameworks, clouds, and messaging channels, with MCP and A2A support.

View on GitHubWebsite
Stars
192
Forks
88
Open issues
112
Last commit
23 hours ago

What is it?

Agent Kernel is a platform layer for running AI agents built with frameworks such as OpenAI Agents SDK, LangGraph, CrewAI, and Google ADK. Developers keep their agent code and let the platform handle deployment, queued request execution, session storage, and knowledge base connections, so switching frameworks or clouds does not require rewriting the application.

The project provides Terraform modules for AWS Lambda and ECS, Azure Functions and Container Apps, and GCP Cloud Run, plus a Helm chart for Kubernetes and on-prem clusters. Requests can run in-process by default or through SQS, Kafka, or NATS JetStream transports. Compliance features include guardrails, PII detection, pre and post execution hooks, and tracing through LangFuse, OpenLLMetry, Pydantic Logfire, or CloudWatch.

Agents can also be exposed to messaging channels such as Slack, WhatsApp, Teams, Telegram, and Gmail, and they support MCP, A2A, and AG-UI protocols. Additional capabilities include sandboxed code execution, scheduled chats, and human-in-the-loop approval pauses.

Key features

  • 01Runs agents from OpenAI Agents SDK, LangGraph, CrewAI, Google ADK, Smolagents, and Pydantic AI side by side.
  • 02Deploys the same agent code to AWS, Azure, GCP, or Kubernetes with included Terraform and Helm assets.
  • 03Connects agents to Slack, WhatsApp, Teams, Telegram, Gmail, Messenger, Instagram, and LiveKit.
  • 04Supports MCP, A2A, and AG-UI protocols.
  • 05Provides guardrails, PII detection, and tracing through LangFuse, Logfire, and CloudWatch.
  • 06Offers pluggable session stores such as Redis, DynamoDB, and Cosmos DB, plus knowledge bases such as ChromaDB and Neo4j.

Use cases

  • Moving a prototype agent built with one framework into a production cloud deployment without rewriting it.
  • Serving the same agent to users across Slack, Teams, WhatsApp, and email.
  • Running compliance-sensitive agents that need guardrails, audit traces, and data residency control.
  • Deploying agents on an existing Kubernetes cluster or in an air-gapped on-prem environment.

Quick start

  1. 1

    Install the package with pip (Python 3.12 to 3.13)

    pip install agentkernel
  2. 2

    Save the multi-agent example from the README as a Python file and run it with the CLI

Things to know

  • Requires Python 3.12 to 3.13, so older or newer Python versions are not supported.
  • The quick start example needs an OpenAI Agents SDK setup, and the README does not show how to supply API keys.
  • Many integrations (cloud deployments, channels, sandboxes) add setup effort beyond the basic install.

In collections

Open-source AI agent frameworks & appsMCP serversOpenTelemetry & observability platforms

Topics

#ai-agents#mcp#a2a#multi-cloud#kubernetes#observability#guardrails#serverless#adk#aws

FAQ

What is agent-kernel?
A Python platform for deploying and running enterprise AI agents across frameworks, clouds, and messaging channels, with MCP and A2A support. Agent Kernel is a platform layer for running AI agents built with frameworks such as OpenAI Agents SDK, LangGraph, CrewAI, and Google ADK. Developers keep their agent code and let the platform handle deployment, queued request execution, session storage, and knowledge base connections, so switching frameworks or clouds does not require rewriting the application.
Is agent-kernel free to use? What license does it have?
agent-kernel is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is agent-kernel still maintained?
Actively maintained. The last commit to agent-kernel was on Nov 6, 2026.
What can I use agent-kernel for?
Moving a prototype agent built with one framework into a production cloud deployment without rewriting it. Serving the same agent to users across Slack, Teams, WhatsApp, and email. Running compliance-sensitive agents that need guardrails, audit traces, and data residency control. Deploying agents on an existing Kubernetes cluster or in an air-gapped on-prem environment.
How do I install agent-kernel?
Install the package with pip (Python 3.12 to 3.13): pip install agentkernel Save the multi-agent example from the README as a Python file and run it with the CLI
Where can I get agent-kernel?
The source code and installation instructions are on GitHub: https://github.com/yaalalabs/agent-kernel

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects