Gigabrain launches 'SuperAgents' in cloud
Gigabrain rolled out 'SuperAgents' — always‑on cloud AI agents with wallets, memory, and real‑time skills — positioning cloud agents as persistent trading assistants and automation layers announced. Persistent agents with state and plug‑in skills change how execution orchestration and market monitoring could be automated across volatile sessions.
Gigabrain is formally sunsetting Scheduled (V1) agents and has disabled creation of new scheduled agents, while stating that existing scheduled agents will continue to operate for the time being (docs.gigabrain.gg). SuperAgents maintain a persistent memory file and the platform documentation specifies the agent itself decides what to remember, calling out trade outcomes, market observations, and lessons as examples of stored items (docs.gigabrain.gg). Each SuperAgent is provisioned with a dedicated Hyperliquid trading wallet, the docs list EVM + Solana support and show funding instructions (send USDC or SOL/ETH for gas), alongside a platform warning to “only send funds you’re willing to risk.” (docs.gigabrain.gg). The feature set supports installable skills built to the Agent Skills open standard, and Gigabrain documents compatibility with Claude Code, OpenClaw, Cursor, GitHub Copilot, Windsurf and Gemini CLI as part of a 26+ platform ecosystem for skills reuse (docs.gigabrain.gg). Gigabrain’s agent runtime describes a deterministic execution cycle — query current wallet/position state, consult memory, analyze market data and microstructure, evaluate strategy instructions, execute or adjust orders, then record the reasoning and outcome to a Learnings datastore (docs.gigabrain.gg). The public changelog and roadmap call out plans to scale to additional exchanges, add Polymarket support, introduce leaderboards and new agent types, while a GitHub agent-skills repository hosts integrations and sample skills for Hyperliquid trading workflows (gigabrain.canny.io).
