AI Agents & Orchestrators

Go-native development harness with 13 safety guardrails (<10ms hooks), Plan→Work→Review→Release workflow, 4-perspective code review, weak-supervision, and rerunnable validation. 1.1K stars, 160 releases.

ByteDance open-source super agent harness. Orchestrates sub-agents, memory, and sandboxes with extensible skills. #1 GitHub Trending. Research, code, build websites, create slides — all autonomously.

Run a team of coding agents on your Mac. Parallel Codex + Claude Code in isolated workspaces, review and merge changes.

Infinite canvas for AI agent orchestration. Connect terminals, draw architecture, scheduled prompts, reusable roles. macOS.

Pre-built AI agents that do the work. 1,000+ integrations, learns preferences, handles emails/schedules/bookkeeping.

Self-improving AI agent by Nous Research. Built-in learning loop, 344 skills, deploys to Telegram/Discord/Slack/WhatsApp.

Open-source AI agent orchestrator. Multi-agent workflows, skill marketplace, community.

Orchestration platform for AI agent companies. Org charts, budgets, governance.

Multi-agent orchestration dashboard for Claude Code. Scoped context containers (tentacles), parallel session management, inter-agent messaging, task delegation via todo.md.

Enterprise agent orchestrator. Connect, deploy, and manage AI agents across your organization with governance controls, compliance, and audit trails.

Claude-native swarm orchestration engine. Deploy 5-20 parallel agents, HNSW + SQLite vector memory, cost tracking, quality gates, CVE scanning.

Microsoft's multi-agent conversation framework.

Multi-agent orchestration framework for collaborative AI teams.

Low-code multi-agent framework. 100+ LLM providers, MCP support, YAML configs, agent handoffs.

Self-improving AI orchestrator. 15 agents, genetic prompt evolution (GEPA), failure pattern synthesis (SIMBA).

Self-evolving skill engine. 46% fewer tokens, agents learn and share skills across sessions.

TypeScript agent framework. Configurable determinism, JSON templates, orchestration steps.

Coding agents as actual teammates. Persistent sessions, memory, async collaboration.

Managed OpenClaw hosting. Deploy AI agents in 60 seconds, 5,200+ agents deployed.

Autonomous bug-bounty framework — 48 agents, 26 commands, 19 CLI tools, 2 MCP servers, 2,047 payload lines. Works across 7 IDEs.

Universal Swarm Intelligence Engine — spawn hundreds of agents that simulate any scenario. Twitter, Reddit, prediction markets, hour-by-hour.

Humans and AI agents on the same page. Shared rooms where Claude, OpenClaw, CrewAI, LangGraph, n8n, and Letta collaborate in real time via WebSocket and REST API.

Framework for building stateful, multi-actor AI apps by LangChain. Graph-based orchestration with cycles, controllability, and persistence.

Stateful LLM agents with long-term memory. Persistent memory, tool use, and multi-step reasoning. Formerly MemGPT.

Conversational AI framework for controllable, context-aware chat agents. Behavior guidelines, glossaries, structured conversations.

Open agent computer — humans and AI agents share browser, files, and apps in one Electron desktop. Continuous memory, roles, templates.

Soul-driven AI agent with permission-hardened tools, token budgets, and multi-channel access. 31 built-in tools, SQLite memory. Runs 24/7 from CLI or Telegram.

Cross-session agent communication via MCP. Claude instances discover and message each other for shared memory and coordinated workflows.

Cloud agent runtime — deploy and manage AI agent clusters on cloud infrastructure. Scaling, monitoring, and orchestration.

Hierarchical Agent Loop Optimization — recursively self-improving harness using RLMs. Collects traces, identifies failure modes, generates fixes automatically.

Self-hosted runtime for autonomous AI agents. Multi-provider, MCP-native, with memory, skills, delegation, schedules, heartbeats, and conversation-to-skill learning. Org chart visualization, desktop app.

Lean, secure personal AI agent built from scratch. Telegram-based, zero exposed ports, user ID whitelist, max iteration limits. No community skill files — MCP only. Inspired by OpenClaw.

AI Chief of Staff framework built on Hermes Agent. Calendar & tasks, social media, research & notes — all through one brain that remembers your context, voice, and priorities. Via MCP → OpenClaw executes.

Run Hermes Agent, OpenClaw, and OpenCode on the same WeChat account. One command install. Switch between AI brains with /hermes, /openclaw, /opencode, /both. Vibe Code via voice messages.

Open-source framework for voice and multimodal AI agents. Real-time conversation pipelines with STT (Groq Whisper), TTS (ElevenLabs, Kokoro), VAD (Silero), video avatars (Pika). Powers War Room-style live agent interactions.

One command to scaffold the complete .claude/ directory. Stack-specific templates (React, Django, Go, Rust, Node, Python, Laravel, Flask), architecture presets (MVP, Production-AWS, Production-GCP), security hooks, IaC, and 100+ template files.

Open-source AI orchestration framework. 12 specialist agents (dev, QA, PM, PO, architect, DevOps, analyst, UX, SM), squad creator, CLI-first workflows.

12 specialized agent squads: Advisory Board (Ray Dalio, Munger), Brand Squad, Hormozi Squad, Cybersecurity, Copy Squad (Ogilvy, Halbert), C-Level, Design, Data, Storytelling, Traffic Masters, Movement, Claude Code Mastery.

Transform Claude Code into a structured development platform. 244K+ downloads, 19 composable skills, project management agents. Works with Claude, Gemini, Qwen.

AI project manager. Feed it requirements and it divides into structured tasks with dependencies. Organize complex multi-step builds.