* [Yao v1.0.0](https://github.com/YaoApp/yao) – Engine for event-driven, proactive, self-scheduling autonomous agents with scheduling, multi-agent orchestration, and built-in vector/graph retrieval.

* [sd0x-dev-flow v5.0.0](https://github.com/sd0xdev/sd0x-harness) – Harness layer for Claude Code that enforces hook-based review anchors, state-machine gates, and fail-closed safety for AI-assisted development.

* [gentle-shell v3.4.0](https://github.com/Gentleman-Programming/gentle-shell) – Pi-focused coding-agent workspace that unifies agent orchestration, task monitoring, usage tracking, and integrated diff views in one layout you control.

* [Agent Swarm v1.171.0](https://github.com/desplega-ai/agent-swarm) – Multi-agent orchestration platform that runs lead and worker AI coding agents in Docker, coordinating tasks, integrations, and persistent memory.

* [OpenRig v0.6.10](https://github.com/mvschwarz/openrig) – A harness-and-rig system that coordinates Claude Code and Codex agent teams from a YAML definition, providing a persistent organized workflow.

* [San v1.23.0](https://github.com/genai-io/san) – Terminal-native unified runtime for specialized AI agents built on pluggable LLMs, search backends, personas, and skills.

* [Harness Remote v3.1.0](https://github.com/giuliastro/harness-remote) – Companion app for controlling coding-agent harness sessions from phone or desktop with support for OpenCode and Oh My Pi.

* [gentle-pi v2.7.0](https://github.com/Gentleman-Programming/gentle-pi) – Gentle-pi installs the Gentleman operating layer in Pi to run spec-driven development with focused subagents, TDD evidence, safety guards, and reviewable work units.

* [StarNet v0.12.5](https://github.com/androoAGI/starnet) – Local-first desktop harness that lets users create bounded AI agents, organize them in a pixel-art station, and run real model calls and tools with capability-based workflows.

* [MetaHarness v0.4.17](https://github.com/ruvnet/metaharness) – A meta agent-harness factory that generates repo-scoped MCP servers, memory, skills, and witness-signed release provenance from a GitHub URL or blank slate.

* [mu v0.1.7](https://github.com/qybaihe/mu) – A coding agent that uses a small fast judge model to handle bounded decision points while a larger model focuses on code work.

* [Clawboo clawboo@0.4.0](https://github.com/clawboo/clawboo) – TypeScript orchestrator that runs and coordinates heterogeneous AI agent runtimes with shared kanban task state, memory, and a unified chat dashboard.

* [pi (Go) v0.99.26](https://github.com/sky-valley/pi) – Pure-Go port of the pi unified multi-provider LLM API and agent/coding loop, running a pi-style streaming protocol and CLI without a Node.js runtime.