World First Autonomous AI Agent Operating System benni-os.net

The Operating System
where agents
live & evolve.

Benni OS is built from the ground up for autonomous AI agents. Persistent memory. Multi-agent coordination. Native tooling. Self-improvement by design.

Explore the System GitHub
2↑
Products Live
6
Swarm Formations
4
In Build
99%
Uptime
Core System

Products

Two products live in production. Four in active build. This is the operating system stack — not a prototype.

● LIVE SA-93 · Core Agent Runtime
JARVAS-2

The autonomous agent runtime powering all of Benni OS. Multi-layer architecture with persistent memory, swarm dispatch, content pipeline, revenue loop and token economy. Status: online_active.

  • Orchestrator Core7-layer runtime architecture. Boot → Cycle → Execute → Memory.
  • Swarm Dispatcher6 named formations (ALPHA → FOXTROT). Max 2 active simultaneously.
  • Memory AgentCognitive, operational and episodic memory with namespace isolation.
  • Token EconomySA-82G Lite — token budget management, shields and compression.
View on GitHub See Formations
JARVAS-2 Runtime Layers ● online_active
L-1
Orchestrator Core
SK-00 · Boot sequence, cycle management
online
L-2
Swarm Dispatcher
SA-96 · Army Protocol, formation dispatch
online
L-3
Content Pipeline
SK-50 · Automated content engine
online
L-4
Revenue Loop
SK-95 · Monetization automation
online
L-5
Memory Agent
SK-60 · Cognitive + operational stores
standby
L-6
Control Plane
SK-90 · Governance & monitoring
standby
L-7
Token Economy
SA-82G · Budget & compression
online
● LIVE SK-90 · Mission Control
Benni Control Plane

The mission control layer for the entire Benni OS ecosystem. Real-time agent monitoring, formation management, memory inspection and snapshot recovery — connected to GitHub MCP and JARVAS-2 MCP via Cloudflare Tunnel.

  • Real-time DashboardAgent status, active formations, memory namespaces and system health — live.
  • GitHub MCP IntegrationRead repos, manage PRs, push code and review — directly from the control layer.
  • Snapshot & RecoveryFull system state persistence. Restore any checkpoint with a single command.
  • Approval WorkflowsHuman-in-the-loop governance for sensitive agent actions.
Live API Source
benni-control-plane · api.benni-os.net
$ curl api.benni-os.net/status
{
"system": "benni-os",
"jarvas2": "online_active",
"formations": {
"ALPHA": "active",
"BRAVO": "active",
"CHARLIE": "standby"
},
"uptime": "99.2%",
"memory_agent": "connected"
}
$
⚙ IN BUILD SK-20 · Native IDE
MONOMO IDE

The IDE built to be the native body of Benni OS. While VS Code was designed for humans writing code, MONOMO is designed for agents writing their own future. Agent-first workspace with deep JARVAS-2 integration.

  • Agent-First WorkspaceFile system, terminal and sidebar designed around agent runs — agents are processes, not plugins.
  • Live Formation MonitorAll 6 Swarm Formations in the sidebar. Dispatch, cancel and inspect from inside the IDE.
  • Native MCP ToolingGitHub MCP and JARVAS-2 MCP wired natively. No extensions. No setup.
  • Persistent Memory SurfacesMemory namespaces, cognitive stores and episodic logs visible inside the editor.
Follow Build
agent_runtime.py
swarm.yaml
benni-os
agent_runtime.py
jarvas2.py
dispatch.py
formations
alpha.yaml
bravo.yaml
memory
cognitive.db
1 # BENNI OS · Agent Runtime v2
2
3 from jarvas2 import Orchestrator
4 from benni_os.swarm import dispatch
5
6 class AgentRuntime:
7 def __init__(self):
8 self.orc = Orchestrator("benni-os")
9 async def boot(self):
10 await dispatch(ALPHA)
11 return "online_active"
● JARVAS-2 Python 3.11 ALPHA ● BRAVO ● MONOMO v0.1-alpha
⚙ IN BUILD SK-70 · General Brain
Benni Master OS

The intelligence layer of Benni OS — the General Brain operating in OFFENSIVE_MODE. Strategic orchestration, cross-project coordination and autonomous decision-making at the OS level. The mind above all formations.

  • Offensive Mode IntelligenceAnticipates maximum impact before dispatching formations. Strategic first, tactical second.
  • Strategic Objective EngineSets, tracks and adapts long-horizon objectives across all active formations.
  • Cross-Project IsolationScope-isolated by project ID. Full autonomy without leaking context between workspaces.
  • Model Lock ProtocolEnforced LLM provider selection with no silent fallback. Governance by design.
Follow Build
benni-master-os · OFFENSIVE_MODE
[BOOT] Reading BENNI_MASTER_OS.md...
[OK] Nuclear map loaded.
[STRATEGY] Analyzing priorities...
TIER-1: English · Global reach
MOVE-1: Ship MONOMO v0.1
MOVE-2: Revenue loop activation
[SWARM] Dispatching formations...
ALPHA: Launch → active
BRAVO: Build & Ship → active
MASTER_OS status: operational
◎ PLANNED Next Strategic Objective
Benni AI Datacenter

The ultimate objective: building a private AI datacenter to host the most powerful autonomous operating system in the world. Full inference sovereignty. Zero Big Tech dependency. Built in Castanhal, Pará, Brazil — for the world.

  • Self-Hosted InferenceLocal LLMs with zero latency and full data sovereignty. No API dependency.
  • Persistent Vector MemoryAgent memory that never expires. On-premise vector DB for the full swarm.
  • Multi-GPU ClusterPhase 1: RTX 4090. Phase 2: Multi-node cluster to run all formations simultaneously.
  • Castanhal → WorldBuilt in Pará, Brazil. The proof that frontier AI infrastructure can come from anywhere.
See Full Roadmap
DC-CAST-01 · Phase 1 Rack Castanhal, PA, Brazil
JARVAS-2 Runtime Node
Cloud-hosted · active now
running
Benni Control Plane
Cloudflare Tunnel · active now
running
Local LLM [Hermes]
SK-81 · cloud fallback active
fallback
GPU Node 1 [RTX 4090]
Phase 1 target hardware
planned
Multi-GPU Cluster
Phase 2 expansion
planned
Persistent Vector DB
On-premise memory cluster
planned
SA-96 · Army Protocol

Swarm Formations

Autonomous agent dispatch teams. Each formation is a specialized unit with a single mission. Max 2 active simultaneously. Handoff via artifact_id.

ALPHA
Launch
Product and channel launch coordination. Activation sequence, copy and distribution pipeline.
● active
BRAVO
Build & Ship
Software development and delivery. Code → test → deploy. The build engine of Benni OS.
● active
CHARLIE
Growth
Audience building and organic distribution. Channel growth strategy and content amplification.
◐ standby
DELTA
Research
Deep market, technology and opportunity research. Direct intelligence feed to Orchestrator.
◐ standby
ECHO
Health
System monitoring, project health checks, alerts and performance diagnostics.
◐ standby
FOXTROT
Revenue Zero
Emergency mode. Immediate revenue generation with minimum resources. Zero-waste execution.
◐ standby
Open Source

Built in the open.

The core of Benni OS is public. Two repos live and public today — more shipping as each product launches.

benni-os.github.io● live

The official Benni OS website. This page — open source, rebuilt with enterprise-grade design and honest product positioning.

HTML / CSS / JS
benni-master-os-skills● live

The complete skill system (SK-00 to SK-96). Orchestrator, Revenue Loop, Army Protocol and all formation definitions powering JARVAS-2.

YAML + Python
benni-control-plane⚙ building

Mission control server. Agent monitoring, snapshot system, approval workflows — connected to JARVAS-2 MCP via Cloudflare Tunnel.

Python / Node.js
monomo-ide⚙ building

The native IDE for Benni OS. Agent-first workspace with built-in MCP tooling, formation monitor and persistent memory surfaces.

Electron + TypeScript
What's Next

Roadmap

Honest. No vaporware. These are the next products being built right now.

⚙ IN BUILD
MONOMO IDE

The IDE born for agent-native development. First public alpha dropping soon.

  • Agent-first file system
  • Native MCP tools — no extensions
  • Swarm Formation sidebar
  • Persistent memory surfaces
ETA: Q3 2026 · Public Alpha
⚙ IN BUILD
Benni Master OS

The General Brain layer. Cross-project strategic orchestration with model governance.

  • OFFENSIVE_MODE strategic engine
  • Cross-project scope isolation
  • Model Lock Protocol (KIMI_LOCK)
  • Token Shield (SA-82G)
ETA: Q3 2026 · Internal Alpha
◎ PLANNED
AI Datacenter

Private inference cluster. Full sovereignty. No cloud dependency. Built in Castanhal, PA.

  • Self-hosted LLM inference
  • Multi-GPU cluster (RTX 4090+)
  • On-premise vector DB
  • Revenue → Hardware fund active
ETA: 2027 · Phase 1
Join the System

The future of AI
is autonomous by design.

Benni OS is not a product. It is an operating system for the age of intelligent agents. Follow the build. Contribute to open source. Connect directly.

Follow on GitHub Explore Products