Skip to main content

Autonomous Orchestration Platform — Signal. Simulate. Execute.

Project description

Orbitaven

Autonomous Orchestration Platform
Signal · Simulate · Execute

PyPI version Python versions License Platform LLM


Orbitaven is a local-first autonomous orchestration platform that watches for signals, builds multi-step workflows, dispatches real AI agents, and executes tasks 24/7 — all through a real-time chat interface running entirely on your own hardware.

No cloud dependency. No SaaS subscription. No data leaving your machine.


Why Orbitaven

Most AI tools wait for you to ask them something. Orbitaven doesn't wait.

It monitors your configured signal sources around the clock. When something happens — a price threshold crossed, a keyword detected, an API returning an unexpected value — it builds a workflow, assigns the right agents, simulates the plan, and executes. You stay in control. The platform does the work. Signal detected → Brain classifies → Workflow generated → Agents dispatched → Result delivered


Key Features

Feature Description
🔁 Signal-driven Monitors URLs, RSS feeds, APIs 24/7. Fires workflows on triggers.
🧠 Brain engine Classifies every input — direct answer or multi-step workflow.
🤖 10 built-in agents file, web, shell, memory, summarizer, coder, data, notify, signal, assistant
🌱 Self-growing Describe a new agent in chat. Orbitaven builds and registers it instantly.
🔐 Credential vault AES-128 encrypted API key storage per profile. Never stored in plaintext.
💬 Context memory Extracts semantic meaning from conversations. Surfaces proactive suggestions.
🤝 Multi-agent negotiation Public protocol — external agents register, submit isolated offers, root agent decides.
Real-time streaming Every workflow step streams output to the UI as it executes.
🌓 Light + dark mode Monochrome sci-fi UI. Fully themeable.
📦 Model-agnostic Ollama (local), Kaggle tunnel, OpenAI, Anthropic, any OpenAI-compatible API.
🖥️ Local-first SQLite, no Docker required, runs on 4GB RAM Ubuntu / Mac / Windows.

Install

Linux / Mac

curl -sSL https://raw.githubusercontent.com/hatsflickcommerce-hub/orbitaven/stage/scripts/install.sh | bash

Windows (PowerShell)

irm https://raw.githubusercontent.com/hatsflickcommerce-hub/orbitaven/stage/scripts/install.ps1 | iex

pip

pip install orbitaven
orbitaven init
orbitaven start

Developers

git clone https://github.com/hatsflickcommerce-hub/orbitaven.git
cd orbitaven
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env
./run.sh

Open browser at http://localhost:8000


Quick Start

# 1. Install
pip install orbitaven

# 2. Initialise (creates config, database, default templates)
orbitaven init

# 3. Start
orbitaven start

# 4. Open browser → http://localhost:8000
# 5. Create a profile → configure your LLM → start chatting

Supported LLM Providers

Provider How
ollama Local Ollama — http://localhost:11434
kaggle_tunnel Kaggle ngrok/cloudflare tunnel URL (free GPU)
openai OpenAI API key
anthropic Anthropic Claude API key
openai_compatible LM Studio, Groq, Together, any OpenAI-spec endpoint

Switch model anytime from Settings panel or by typing in chat.


Built-in Agent Pool

Agent Capability
assistant General LLM reasoning, writing, answering
file_agent Read, write, list, search local files
web_agent Browse URLs, DuckDuckGo search (no API key)
shell_agent Safe whitelisted shell commands
memory_agent Persistent per-profile SQLite memory
summarizer_agent LLM-powered content condensing
coder_agent Code generation + sandboxed Python execution
data_agent CSV/JSON parsing, calculations, statistics
notify_agent Alerts, notification log
signal_agent URL/RSS/API monitoring

Grow your own

You: "Create an agent that fetches my GitHub issues" Orbitaven: "What's your GitHub token? It will be stored encrypted." You: "ghp_xxxx" Orbitaven: ✅ github_agent created and registered.


Chat Commands

Command Description
/memory Show what Orbitaven remembers about you
/suggest Get proactive action suggestions from your context
/signals Show all configured signals and active alerts
/agents List all available agents
/help Show all commands

Multi-Agent Negotiation Protocol (Phase 8)

External developers can register agents and participate in negotiation rounds:

# Register your agent (get API key)
POST /api/negotiate/register
{"name": "MyAgent", "description": "...", "capabilities": [...]}

# Submit offer to open session (isolated — never sees other offers)
POST /api/negotiate/offer
X-API-Key: oa_xxxx
{"session_id": "...", "offer": {"price": 5000, "delivery": "3 days", ...}}

# Host triggers root agent evaluation → winner announced
POST /api/negotiate/sessions/{id}/evaluate

Information isolation guaranteed — agents never see each other's offers.


CLI Reference

orbitaven init              # initialise config and database
orbitaven start             # start platform (foreground)
orbitaven start --daemon    # start in background
orbitaven start --reload    # hot reload (development)
orbitaven start --port 9000 # custom port
orbitaven stop              # stop background instance
orbitaven status            # check if running
orbitaven version           # show version

Architecture

orbitaven/ ├── src/orbitaven/ │ ├── core/ │ │ ├── brain.py — intent classifier + workflow generator │ │ ├── agent_bus.py — inter-agent pub/sub communication │ │ ├── context_engine.py — semantic memory extraction │ │ ├── negotiation_engine.py — multi-agent negotiation protocol │ │ ├── signal_monitor.py — 24/7 background daemon │ │ ├── workflow_manager.py — state machine │ │ └── credential_vault.py — AES encrypted secrets │ ├── agents/ — 10 built-in + generated agents │ ├── api/ — FastAPI + WebSocket │ └── ui/ — Monochrome sci-fi single-page app ├── scripts/ │ ├── install.sh — Linux/Mac installer │ └── install.ps1 — Windows installer └── pyproject.toml — pip packaging


Hardware Requirements

Component Minimum
OS Ubuntu 20.04+ / Mac / Windows
RAM 4GB
Storage 10GB
Python 3.11+
GPU Not required (use Kaggle tunnel for free GPU inference)

Build Phases

Phase Feature Status
1 Core — config, DB, profiles, LLM, WebSocket, UI
2 Brain engine — intent routing, workflow state machine
3 Agent pool — 10 built-in agents, real execution
4 Signal system — 24/7 monitor, alerts, one-click launch
5 Agent factory — self-growing agents, credential vault
6 Polish — health dashboard, run.sh, system stats
7 Intelligence — AgentBus, ContextEngine, suggestions, light mode
8 Negotiation — public multi-agent protocol, isolation
9 Cloud — Docker, Telegram/WhatsApp, mobile API 🔜

Contributing

git clone https://github.com/hatsflickcommerce-hub/orbitaven.git
cd orbitaven
python3 -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"

Issues and PRs welcome at github.com/hatsflickcommerce-hub/orbitaven


License

MIT © 2026 Orbitaven


Signal · Simulate · Execute

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

orbitaven-0.3.1.tar.gz (83.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

orbitaven-0.3.1-py3-none-any.whl (98.7 kB view details)

Uploaded Python 3

File details

Details for the file orbitaven-0.3.1.tar.gz.

File metadata

  • Download URL: orbitaven-0.3.1.tar.gz
  • Upload date:
  • Size: 83.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.4

File hashes

Hashes for orbitaven-0.3.1.tar.gz
Algorithm Hash digest
SHA256 c71da21174e0c2e5a638a02123ef922d46ae5a479936b02f052150f83ae75d8a
MD5 b87c0a03cd4af31d572b769ec5380962
BLAKE2b-256 39db7e1d792412376f2d16762c83972dee52555b1c3a50aa6dde8287163516be

See more details on using hashes here.

File details

Details for the file orbitaven-0.3.1-py3-none-any.whl.

File metadata

  • Download URL: orbitaven-0.3.1-py3-none-any.whl
  • Upload date:
  • Size: 98.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.14.4

File hashes

Hashes for orbitaven-0.3.1-py3-none-any.whl
Algorithm Hash digest
SHA256 bfec5f83db9aa7f324be209c4f3820c1c82b1dbab0a2d273914188ffb88fd6a5
MD5 a3dd28012ffb054da27b2ee2d11efb1f
BLAKE2b-256 4afd2ccb38592b75afe1dab03999980ad4f91bba7ba8df973786eb2b3eb69308

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page