mahimalam/topological-convergence-router — reverse-engineered prompt

Build me a Python 3.10 daemon called Topological Convergence Router. I want it to watch a live stream of simulated event nodes, keep their current probability states in memory, and constantly scan the graph for groups of connected nodes where the math no longer balances, like when the combined probability drops below 1.0.

When it finds a divergence, it should use Depth First Search to map the path, calculate the edge percentage, create a resolver payload, and send it into an async priority queue so the highest value opportunities get handled first. Please include a late state scanner that polls volatile nodes more aggressively near their unlock time.

Keep it safe and local only, with mock data instead of real network calls, secrets, or proprietary execution logic. Add clear config values for thresholds and limits, logging so I can see what the daemon is doing, and a simple way to run a demo from the terminal.