Open Source · Apache 2.0
CyberneticAgents is a terminal-based research system where 12 LLM-powered agents collaborate through analysis, debate, and risk assessment — then generate autonomous trading agents that predict, track performance, and improve over time.
Star the repo on GitHub
Features
LangGraph orchestration, Rich terminal UI, SQLite-backed paper trading, and support for six LLM providers — all running locally in your terminal.
Research pipeline
Run a full multi-agent analysis on any ticker. Market, fundamentals, news, and social analysts gather data independently. Bull and bear researchers debate. A risk panel stress-tests the thesis. A trader and risk manager deliver a final BUY/SELL/HOLD.
Agent creation
Generate persistent trading agents from deep research or describe a strategy in natural language. Each agent gets a unique persona, custom system prompts, tool configuration, and a $10,000 paper portfolio to track performance.
Lightweight debate
First runs use a single LLM call. On repeat runs, stale research triggers tiered context refresh — fresh news, sentiment, prediction history — then a 3-agent debate replaces the single prediction. Smarter updates without rerunning the full pipeline.
Analysts, researchers, risk debators, and traders collaborate through structured phases with LangGraph orchestration.
Automatic staleness detection enriches agents with fresh news, sentiment, and prediction history on repeat runs.
SQLite-backed predictions with automatic resolution, P&L calculation, and portfolio management — all local.
yfinance by default with Alpha Vantage support and per-category vendor routing for prices, fundamentals, and news.
Interactive menus, live dashboard, formatted reports, animated ASCII banner, and Deep Space / Neon Pulse themes.
Run cybernetic.biz update to pull the latest version from GitHub. One install script gets you running.
How it works
Research produces the knowledge. Agents act on it repeatedly — with tiered refresh and lightweight debate keeping predictions current without rerunning the full pipeline every time.
Market · Fundamentals · News · Social
Up to four analyst agents independently gather data using specialized tools and produce detailed reports.
Bull · Bear · Research Manager
Bull and bear researchers debate the investment case over configurable rounds. A research manager judges and produces an investment plan.
Trader
A trader agent takes the investment plan and formulates a concrete strategy with entry/exit points and position sizing.
Aggressive · Neutral · Conservative · Risk Manager
Three risk debators argue the profile. A risk manager synthesizes their debate into a final BUY/SELL/HOLD decision.
$ cybernetic.biz analyze
Phase 1 Market Analyst · Fundamentals · News · Social
Phase 2 Bull Researcher ↔ Bear Researcher → Research Manager
Phase 3 Trader → Trading Strategy
Phase 4 Risk Debators → Risk Manager → BUY / SELL / HOLDQuick start
Requires Python 3.10+ and at least one LLM provider API key. The install script sets up a virtual environment, installs dependencies, and symlinks cybernetic.biz to your PATH.
cybernetic.biz updategit clone https://github.com/CYBERNETIC-BIZ/CyberneticAgents.git
cd CyberneticAgents
./install.sh$ cybernetic.bizLLM Providers
Switch providers through the interactive menu when researching a ticker or creating an agent. Deep think and quick think models are configurable separately.
GPT-5.2, GPT-5-mini, GPT-4.1
Claude Sonnet 4.5, Opus 4.5, Haiku 4.5
Gemini 3 Pro/Flash, Gemini 2.5 Flash
Grok 4.1, Grok 4
Any model via OpenRouter
Auto-detected local models