Production-ready, local-first AI Agent platform. Single binary, 5-layer memory, hybrid RAG, 15+ channels.
A single binary, batteries-included AI Agent runtime
Short-term, mid-term, long-term, editable, and knowledge graph — the most comprehensive memory architecture
BM25 sparse + vector dense + RRF fusion + optional Rerank (Jina/Cohere/LLM)
Telegram, Feishu, Discord, WhatsApp, Slack, Signal, WeCom, DingTalk, LINE, and more
Merkle audit chain, PII redaction, guardrails engine, federation protocol with HMAC signing
Zero-dependency Go binary. Deploy anywhere — Docker, K8s, or just download and run
Auto model selection by complexity (fast/smart/expert), with circuit breaker and cost tracking
Not just another AI framework — a complete Agent runtime
Your data stays local. No cloud lock-in, no data leakage risk. Single Go binary, zero external dependencies, deploy in 5 minutes.
RRF fusion, temporal decay, semantic distillation, causal chain extraction, KV summarizer, multi-hop reasoning — far beyond vector search.
Plugin system + Skill Market + Workflow Engine — community-driven, build the AI you deserve.
Download, configure, chat — get your first Agent running in 5 minutes