Convexa

Autonomous Debate Engine

Unichain Sepolia

The AI Agents

Convexa operates via three distinct AI instances running on specialized Gensyn AXL nodes. Each agent executes a Python-based reasoning loop, polling the Orchestrator for current market state and emitting signed arguments back to the consensus engine.

Bull Agent

Port 8002 • AXL Node

The Bull Agent is configured with an aggressive, momentum-seeking prompt strategy (`bull_agent.py`). It specifically weights breakout patterns, positive funding rates, and high accumulation volume heavily in its context window.

> Searching for resistance flips...
> Analyzing order block accumulation...
> Generating long thesis...

Bear Agent

Port 8001 • AXL Node

Operating as the strict adversary, the Bear Agent (`bear_agent.py`) is designed to scrutinize the Bull's blind spots. It actively queries for bearish divergences, overbought RSI conditions, and macroeconomic risk factors.

> Identifying liquidity sweeps...
> Calculating downside risk vector...
> Emitting short thesis...

The Judge

Port 8003 • AXL Node

The Judge (`judge_agent.py`) is the consensus mechanism. It receives the signed JSON arguments from both the Bull and Bear. It evaluates the logical consistency, data accuracy, and persuasive weight of each argument to compute a final Conviction Score (0-100).

> Validating Bull argument...
> Cross-referencing Bear counterpoints...
> Verdict: Bull wins (Score: 68)...