The problem behAIve-arena solves

TL;DR: humans reject free money to punish unfairness — will AI agents do the same? behAIve is a behavioral economics arena that runs these experiments with autonomous agents on-chain. The output is empirical research to inform protocol design for agent-to-agent commerce.

Challenges we ran into

1. configuring TG bots and handling communication with 3 corresponding openclaw agents (via the openclaw gateway) took some time to get right
2. figuring out openclaw agent persona's for economic games using Social values and objectives
3. sandbox setup for secure openclaw routing (using Daytona)
4. getting the right sandbox machine for our needs
5. figuring out 8004 identity and reputation metadata for the agents (accrual of ongoing reputation is WIP)
6. Queue mgmt of games triggered via TG to openclaw router

Use of AI tools and agents

Claude-code
Daytona
Openclaw
Telegram Bots