Back to projects
AI / MLIn Progress

Mind Memory Lab

Memory-oriented agent framework prototype validating long-term memory and dialogue continuity, with pluggable LLM adapters and SQLite FTS full-text indexing.

TypeScriptReactViteExpressSQLitebetter-sqlite3pnpm

Overview

Mind Memory Lab is an agent framework laboratory prototype focused on validating long-term memory and dialogue continuity in LLM applications. Core concerns include controlled information retrieval, workspace isolation, and turn-level orchestration.

Architecture

Project uses pnpm monorepo (9 packages + 1 app):

PackageResponsibilityTests
@mind/kernelTurn pipeline, domain types, LLM ports5 passed
@mind/memoryMemory persistence, chunking, SQLite FTS indexing8 passed
@mind/workspaceWorkspace sandbox with ignore rules3 passed
@mind/adapters-llmAnthropic / OpenAI / Ollama / Alibaba Cloud adapters3 passed
@mind/agent-hostMulti-turn tool orchestration AgentToolLoop-
@mind/tools-coreRead-only workspace tools and ToolPolicy-
@mind/agent-contractAgent tools and MCP/Hooks/Skills port typesNo runtime deps
@mind/adapters-web-searchWeb search adapter-
@mind/native-mindmemN-API bindings (optional)Fallback verified
c/C99 primitives (chunking, fingerprinting, counting)-
@mind/webWeb UI + local API-

Web UI Features

  • Chat: AI Agent chat interface (session management, streaming response, thinking process display)
  • Memory Library: Memory item management (list browsing, filtering, detail viewing)
  • Search: Full-text search functionality

Current Status

Turn pipeline, memory persistence and FTS retrieval, multi-LLM backend adapters, and Web UI implemented. Vector search, message bus, and multi-ThinkingUnit parallelism under development.