Useful projects. Clear sources.
Find AI, agents and token-powered tools by what they actually do.
Use a project name, capability or a task in your own words.
Adapt PyTorch training scripts to multiple devices and mixed precision with a shared execution interface.
The README demonstrates changes to a training loop and explains single-device, multi-device and mixed-precision execution.
Build application automations and agent tools using reusable integration pieces and approval steps.
The official README documents TypeScript integration pieces, a visual builder, MCP exposure and approval or delay steps.
Connect agent backends to user interfaces through a shared stream of interaction events.
The official README specifies event-based agent-to-application interaction and links reference transport and framework implementations.
Provide a configurable chat frontend for a LangGraph server's message-based agents.
The official README documents server connection settings, assistant selection, authentication configuration and chat application setup.
Develop computer-use agents that interact with desktop applications through screen, mouse and keyboard controls.
The official repository describes its GUI agent and provides implementation, setup and research links for computer interaction.
Run an agent workspace with a Linux desktop, browser, documents, skills and plugins.
The official README describes the Docker-based workspace, desktop launcher, browser support and configuration workflow.
A communication protocol for independent agents to discover capabilities, delegate tasks and exchange results.
Official documentation includes agent discovery, task lifecycle, specifications and SDK links.
Build agents from tools, model interfaces and workflow components, with console and web service entry points.
The official README includes an agent construction example, service deployment guidance and A2A and pipeline documentation links.
Compose domain-focused agents using collaboration patterns such as planning, execution and review.
The official README describes reusable collaboration components and the Plan, Execute, Express and Review pattern.
Build and serve agents with stored sessions, tool integrations, shared memory and approval steps.
The official README documents the SDK, AgentOS runtime, database storage, toolkits and approval controls.
Pair-program from a terminal with repository context, editable diffs, Git commits and test feedback.
The official README describes repository maps, source editing, Git integration, linting and testing, with installation documentation.
A decentralized cloud marketplace with deployment tools for applications and compute workloads.
Official documentation supplies deployment guides, provider operations and SDK/API references.
A growing index of projects with practical utility. Reviewed sources document a project’s purpose; community listings describe themselves. Listings do not verify every claim or imply a partnership.
