{"id":"e33bd8c4-827e-47df-99f6-e1a4038201e2","slug":"hyper-agent-os","kind":"mcp","name":"hyper-agent-os","summary":"Next-Generation Distributed Multi-Agent Runtime & Swarm Substrate uniting Super-Memory, Cyclical Swarms, 24/7 Durable Daemons, Multimodal Streaming, VLA Safety Governors, and Benchmark Evaluators.","owner":"AAH20","repoUrl":"https://github.com/AAH20/hyper-agent-os","license":"Apache-2.0","licenseTier":"permissive","trust":"community","stars":0,"match":null,"fewStars":true,"format":"mcp","createdAt":"2026-09-18T16:02:15.000Z","addedAt":"2026-10-21T06:50:54.853Z","logoUrl":"https://avatars.githubusercontent.com/u/51206155?u=2d57ecbbaa9391ec6cf4ca6fc44a324fc7fe591c&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["memory-knowledge","search","data-analytics"],"tags":["ai-agents","autonomous-agents","control-barrier-functions","crewai","devin-alternative","durable-execution","graphrag","knowledge-graph"],"runtime":"python","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-26T11:19:13.283Z","sample":false,"url":"https://conduits.dev/tools/hyper-agent-os","language":"Python","transports":["stdio"],"install":{"pypi":"hyper-agent-os"},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# ⚡ Hyper-Agent OS: Distributed Multi-Agent Runtime & Swarm Substrate\n\n<div align=\"center\">\n\n[![License: Apache-2.0](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)\n[![Python Version](https://img.shields.io/badge/Python-3.10%20%7C%203.11%20%7C%203.12%20%7C%203.14-blue?logo=python&logoColor=white)](https://python.org)\n[![Tests](https://img.shields.io/badge/Tests-29%2F29%20Passing%20(100%25)-success?logo=pytest)](file:///Users/ahmedhassan/Downloads/a2z-soc-main%202/hyper-agent-os/tests)\n[![MCP Compliant](https://img.shields.io/badge/MCP-Model%20Context%20Protocol-8A2BE2?logo=anthropic)](https://modelcontextprotocol.io)\n[![SWE-bench Verified](https://img.shields.io/badge/SWE--bench-Verified%20%26%20Multimodal-orange)](https://www.swebench.com)\n[![MLE-bench Kaggle](https://img.shields.io/badge/MLE--bench-Kaggle%20Medal%20Grade-20BEFF?logo=kaggle)](https://github.com/openai/mle-bench)\n[![Post-Quantum Attestation](https://img.shields.io/badge/Aegis-Post--Quantum%20Audit%20Chains-red)](#5-critical-systems--vla-safety-governor)\n\n**The unified, enterprise-grade distributed agent runtime uniting Super-Memory (Bi-Temporal GraphRAG), Cyclical Swarm Orchestration, 24/7 Durable Daemons, Multimodal Streaming, and Critical Systems VLA Safety Governors.**\n\n[Key Features](#-key-features) • [Competitive Matrix](#-competitive-feature-matrix) • [Quickstart](#-quickstart-in-30-seconds) • [Architecture](#-architecture) • [MCP Server Setup](#-model-context-protocol-mcp-server) • [Benchmarks](#-benchmark-evaluators)\n\n</div>\n\n---\n\n## ⚡ Why Hyper-Agent OS?\n\nMost AI agent frameworks are either:\n1. **Stateless prompt chains** that hallucinate when context grows (naive vector RAG).\n2. **Fragile Python loops** that crash, drop state, or leak memory during 24/7 background execution.\n3. **Unconstrained probabilistic actors** dangerous to deploy to physical actuators, robotics, or critical production infrastructure.\n\n**Hyper-Agent OS** solves this by unifying a **Bi-Temporal Graph Memory** (surpassing Cognee), **Cyclical State Graphs** (LangGraph-style) with **Byzantine Quorums**, **Temporal-grade Durable Execution**, **Sub-100ms Multimodal Streaming**, and **Deterministic Control Barrier Functions (CBF)**.\n\n---\n\n## 📊 Competitive Feature Matrix\n\n| Feature | Hyper-Agent OS | Cognee | LangGraph | CrewAI | Naive Vector RAG |\n| :--- | :---: | :---: | :---: | :---: | :---: |\n| **Bi-Temporal Knowledge Graph** | **✅ Yes** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **Point-in-Time Rollback (Anti-Hallucination)** | **✅ Yes** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **3D Spatial & Multimodal Embeddings** | **✅ Yes** | ❌ No | ❌ No | ❌ No | ⚠️ Flat vectors |\n| **Cyclical State Graphs & Interrupts** | **✅ Built-in** | ❌ No | ✅ Yes | ❌ Linear | ❌ No |\n| **Byzantine Fault Tolerant Swarm Quorum** | **✅ BFT Quorum** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **24/7 Durable Replay (Temporal Pattern)** | **✅ SQLite Journal** | ❌ No | ⚠️ Checkpoints | ❌ No | ❌ No |\n| **Sub-100ms Voice Stream + Barge-In** | **✅ Yes** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **Critical Safety Governors (CBF)** | **✅ Deterministic** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **Post-Quantum Attestation Receipts** | **✅ Aegis/SHA-256** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **Built-in SWE-bench & MLE-bench Eval** | **✅ Yes** | ❌ No | ❌ No | ❌ No | ❌ No |\n| **Model Context Protocol (MCP) Server** | **✅ Native** | ⚠️ Plugins | ⚠️ Community | ⚠️ Community | ❌ No |\n| **Drop-in Cognee Migration Shim** | **✅ 1-Line Import** | N/A | ❌ No | ❌ No | ❌ No |\n\n---\n\n## 🏛️ Architecture\n\n```\n                               ┌────────────────────────────────────────┐\n                               │       Hyper-Agent OS / Runtime         │\n                               └───────────────────┬────────────────────┘\n                                                   │\n     ┌──────────────────────┬──────────────────────┼──────────────────────┬──────────────────────┐\n     ▼                      ▼                      ▼                      ▼                      ▼\n[ 1. Memory Substrate ]  [ 2. Swarm Core ]  [ 3. 24/7 Daemon ]  [ 4. Streaming ]      [ 5. Safety & VLA ]\n• Bi-Temporal Graph      • LangGraph DAG    • Durable Loops     • WebRTC/Audio Stream • Control Barrier Func\n• Vector Index           • CrewAI Swarms    • Event Bus (PubSub)• Video Keyframe Pipe • Actuation Envelope\n• Spatial Scene Graph    • Role Consensus   • Self-Healing      • Real-Time VAD/TTS   • Deterministic Reject\n                                                   │\n                                                   ▼\n                                  [ 6. Benchmark Evaluation Suite ]\n                                  • SWE-bench Verified & Multimodal Runner\n                                  • MLE-bench (Kaggle) Evaluation Adapter\n                                  • ARC Reasoning & Tool Execution Harness\n```\n\n---\n\n## 🚀 Quickstart in 30 Seconds\n\n### Installation\n```bash\ngit clone https://github.com/AAH20/hyper-agent-os.git\ncd hyper-agent-os\npip install -e .\n```\n\n### Run the CLI Demonstration\n```bash\n# Run live end-to-end demonstrations across all 6 subsystems\nhyper-os all\n\n# Or run individual subsystems\nhyper-os memory\nhyper-os swarm\nhyper-os daemon\nhyper-os streaming\nhyper-os safety\nhyper-os benchmark\n```\n\n### Run the Automated Test Suite (100% Passing)\n```bash\npython3 -m unittest discover -s tests -p \"test_*.py\" -v\n```\n\n---\n\n## 🔄 Drop-in Cognee Migration\n\nMigrate from Cognee in one line of code:\n\n```python\nfrom hyper_agent_os.adapters import CogneeCompat\n\ncognee = CogneeCompat()\n\n# Ingest unstructured text or code\ncognee.add_sync([\n    \"QuantumController coordinates RobotArmAlpha.\",\n    \"RobotArmAlpha is governed by ControlBarrierFunction.\"\n])\n\n# Extract and cognify into the bi-temporal graph\ncognee.cognify_sync()\n\n# Multi-hop hybrid graph retrieval\nresults = cognee.search_sync(\"What governs RobotArmAlpha?\", subject=\"RobotArmAlpha\")\nprint(results)\n```\n\n---\n\n## 🔌 Model Context Protocol (MCP) Server\n\nConnect Hyper-Agent OS directly to **Cursor, Windsurf, Claude Desktop, or Antigravity**:\n\n### Claude Desktop / Cursor Config (`mcpServers`):\n```json\n{\n  \"mcpServers\": {\n    \"hyper-agent-os\": {\n      \"command\": \"python3\",\n      \"args\": [\"-m\", \"hyper_agent_os.mcp_server\"],\n      \"env\": {}\n    }\n  }\n}\n```\n\n### Native Tools Exposed via MCP:\n- `hyper_query_memory`: Query the Bi-Temporal Knowledge Graph & Multimodal index.\n- `hyper_assert_fact`: Commit persistent facts with cryptographic audit attribution.\n- `hyper_run_swarm`: Launch autonomous collaborative swarm workflows.\n- `hyper_evaluate_cbf_safety`: Certify and project actuator commands onto safe operational envelopes.\n\n---\n\n## 🐳 Docker & 24/7 Production Deployment\n\nRun Hyper-Agent OS as an autonomous background daemon with persistent SQLite journaling:\n\n```bash\ndocker compose up -d\n```\n\n---\n\n## 📈 Benchmark Evaluators\n\nHyper-Agent OS includes built-in harnesses for the industry's most demanding AI benchmarks:\n\n1. **SWE-bench Verified & Multimodal**:\n   - Tests patch generation against real GitHub repository issues.\n   - Evaluates whether unit test failures flip to pass (`FAIL_TO_PASS`) without regressing existing tests.\n2. **MLE-bench (OpenAI / Kaggle)**:\n   - Evaluates end-to-end Machine Learning pipelines on Kaggle competitions.\n   - Automatically grades models into **Bronze, Silver, or Gold Medal tiers**.\n3. **ARC Prize (Abstraction and Reasoning Corpus)**:\n   - Measures out-of-distribution reasoning and visual grid program synthesis.\n\n---\n\n## 🏷️ Recommended GitHub Topics (SEO Tags)\n\nAdd these exact topics in your GitHub repository settings to maximize search indexing:\n\n```text\nai-agents, graphrag, knowledge-graph, bi-temporal-memory, langgraph, \ncrewai, swarm-intelligence, mcp, model-context-protocol, swe-bench, \nmle-bench, autonomous-agents, durable-execution, voice-agent, vla, \nrobotics, control-barrier-functions, post-quantum, devin-alternative, rag\n```\n\n---\n\n## 📜 License & Authors\nDeveloped by **Ahmed Hassan** (Founder, A2Z SOC).  \nLicensed under the [Apache-2.0 License](LICENSE).\n"}