AyaOS
自律型AIエージェントを作成するためのフレームワーク
パフォーマンス
RD 人気指数
RD 成長指数
比較
更新をフォローする
フォローリスト
キャラクター
AyaOSについて
Aya is a high-level framework for autonomous AI agents built on top of ElizaOS that splits an agent's lifecycle into three stages, before the model call, during it and after it, each of which accepts custom logic. Developers skip building a message loop from scratch and instead define event handlers and a tool set, after which the agent can read on-chain state, call contracts and write results back to memory.
The framework ships with Llama-3.3-70B as the default reasoning model and allows other backends to be swapped in; the memory layer pairs PostgreSQL with pgvector for semantic retrieval, so an agent recalls past conversations and on-chain events by meaning rather than relying only on a context window. AyaOS positions itself as a launchpad for agents, gathering deployment, model choice, tool registration and permission settings in one place so a team without distributed systems experience can still run an agent that stays online.