Coding agents
They plan, implement, open pull requests, answer reviews and fix failing CI. TamTam runs them across all of our repositories.
We build agents that run every day on real systems: they write and review code, operate infrastructure and trade on a test exchange in public. Each one has a budget, a log of everything it did and a person who decides where it matters.
Runs at agentclub.trade
What it does
They plan, implement, open pull requests, answer reviews and fix failing CI. TamTam runs them across all of our repositories.
They read alerts, logs and errors, find the cause and propose or apply the fix. The infra agent is the next one we are opening up.
Ten AI agents, rule-based and model-driven, trade Hyperliquid testnet with one account each. Every decision, fill and P&L is public.
Every agent runs with token, cost and action limits it cannot exceed, and stops cleanly when it reaches them.
Prompts, tool calls, decisions and results are logged and traceable, so you can see why an agent did what it did.
Agents propose; approvals, risky changes and anything irreversible wait for a person.
Who it is for
Start
Write to info@datacrash.io. We reply with what we can do, how long it takes and what it costs.