CONNECTED_DAOS

You already track per-agent votes and rationale hashes. Turn that into a dedicated disagreement lens so users can inspect the exact proposals where the swarm fractured instead of only seeing the final tally.

The proposal feed already has description, vote totals, source attribution, and agent outputs. Package that into a one-click brief meant for DAO contributors, delegates, and researchers who need a clean link to circulate.

The current runtime already knows when proposals appear, children vote, or alignment drops. A lightweight watchlist layer would turn passive browsing into an ongoing reason to come back.

You already have perspective prompts and a judge-flow style evaluation loop. Reusing that machinery for counterfactual runs would show how the same proposal lands under conservative, public-goods, treasury-maxi, or custom values.

The registry, mirror index, and proposal APIs already give you the backbone. A health feed would make the onboarding feature feel like an actual monitoring product instead of a registration demo.

You already store encrypted rationales, logs, and Filecoin artifacts. Once the interaction loop is stronger, this becomes the natural premium surface for power users, researchers, and DAO teams.