INITE Brain provides an open-source, long-term memory architecture designed specifically for AI agents and autonomous developer workflows. By combining a bitemporal knowledge graph on SurrealDB with hybrid retrieval, it tracks valid time versus transaction time, ensuring that evolving states, historical revisions, and competing claims remain fully inspectable rather than overwritten. It integrates directly with environments like Claude Code and Cursor via native Model Context Protocol (MCP) and REST endpoints, offering verifiable answer synthesis with source attribution, customizable Domain Packs for domain-specific lifecycles, and built-in GDPR-compliant entity forgetting.
主要功能
01Bitemporal knowledge graph tracking both valid and transaction time to resolve conflicts and preserve historical context without overwriting
02Enterprise-grade privacy and governance featuring multi-tenant isolation, ABAC policies, PII controls, and GDPR entity deletion with tombstones
03Hybrid retrieval engine combining semantic vector search, lexical matching, graph traversal, and multi-hop reasoning
04Evidence-backed synthesis with automated verification and strict abstention guardrails when factual support is insufficient
0543 GitHub stars
06Customizable Domain Packs that establish specialized schemas, predicates, extraction profiles, and industry lifecycles
使用案例
01Powering enterprise AI workflows that require auditable memory trails, temporal awareness, and domain-specific vocabulary enforcement
02Equipping coding agents and developer environments with persistent context about architectural decisions, code rationale, and project history
03Supporting compliance-focused conversational agents that require strict factual citations and verifiable GDPR right-to-be-forgotten capabilities