Skip to content
Agent Search Engine.
Advertise
Menu

Open source · Python

Best Python agent infrastructure

The 11 open-source agent infrastructure written in Python in this index, ordered by recorded GitHub star count, never by sponsorship. Stars are repository-interest signals, not performance or maintenance guarantees. mem0 leads the set at 67,437 GitHub stars, and 8 of the 11 have a recorded repository push in the last 90 days. These records have Python as their primary repository language and recognized open-source software licenses recorded in this index; check the current licenses, deployment requirements, and model or API costs. Right now it's led by mem0 (67k stars), with Cognee and Graphiti following. Order follows the recorded star counts; repository activity is shown separately.

11 open-source records · all agent infrastructure · more best-of lists · how we rank

mem0InfrastructureUniversal memory layer for AI AgentsOpen sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 67kGitHub starsCogneeInfrastructureAn open-source memory platform that turns documents, code and conversations into a knowledge graph for agents.Open sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 33kGitHub starsGraphitiInfrastructureA framework for building temporal context graphs that give AI agents memory of how facts change.Open sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 32kGitHub starscuaInfrastructureOpen-source infrastructure for Computer-Use Agents. Sandboxes, SDKs, and benchmarks to train and evaluate AI agents that can control full desktops (macOS, Linux, Windows).Open sourceCosts not checkedPython · recorded license: MITRecord checked · Recent push 30kGitHub starsOpikInfrastructureComet's open-source platform for LLM and agent tracing, evaluation, prompt management and monitoring.Open sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 23kGitHub starsE2BInfrastructureOpen-source infrastructure for running AI-generated code in secure, isolated cloud sandboxes.Open sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 14kGitHub starsAIO SandboxInfrastructureAn all-in-one Docker sandbox for AI agents with a browser, shell, files, VS Code server, Jupyter and MCP.Open sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 6.2kGitHub starsklavisInfrastructureKlavis AI: MCP integration platforms that let AI agents use tools reliably at any scaleOpen sourceCosts not checkedPython · recorded license: Apache-2.0Record checked 5.8kGitHub starsmcpoInfrastructureA simple, secure MCP-to-OpenAPI proxy serverOpen sourceCosts not checkedPython · recorded license: MITRecord checked 4.4kGitHub starsHOL GuardInfrastructureLocal policy checks, approval prompts and action receipts for supported agent tools.Open sourceCosts not checkedPython · recorded license: Apache-2.0Record checked · Recent push 934GitHub starsMCP-BridgeInfrastructureA middleware to provide an openAI compatible endpoint that can call MCP toolsOpen sourceCosts not checkedPython · recorded license: MITRecord checked 923GitHub stars

Frequently asked

What are the best Python agent infrastructure?
By recorded GitHub star count, mem0, Cognee and Graphiti lead the Python agent infrastructure in this index of 11. The full list is below; star count does not prove suitability or continued maintenance. Sponsorship does not affect its order.
Are these Python agent infrastructure open source and free?
These records have recognized open-source software licenses recorded and are written primarily in Python in this index. Apache-2.0 is the most common licence here, on 8 of 11, across 2 licences in total. Check each license and deployment requirements. Hosting, model, API, and managed-service costs may apply.
Why choose Python agent infrastructure specifically?
Staying in your team's primary language — Python — can help your team inspect and modify code it understands. Repository language does not establish supported user languages, SDKs or an offline deployment. 8 of these 11 have a recorded push within 90 days. A quiet repository alone does not establish whether the project is complete, supported or abandoned.