Victor Memory Hub

Victor Memory Hub

Four-tier memory architecture with automated Dreaming 流水线, three-way 同步hronization, Active Memory recall, and two-stage auto-提取ion.

Overview Tier Layer Techno记录y Purpose L0 运行time Retrieval memory-core 插件 (Ollama bge-m3 → SQLite + sqlite-vec) Real-time semantic + BM25 hybrid 搜索 L0 Cloud Recall MemOS Cloud 插件 (optional) Cross-device memory capture and recall L0 Active Memory Built-in OpenClaw 插件 Pre-reply sub-代理 memory 搜索 + 上下文 injection L1 Working Memory memory/YYYY-MM-DD.md files DAIly summaries, todos, technical notes L2a Long-term 索引 MEMORY.md (~90 lines, read-only base) Bidirectional 索引 → 归档.md, 0% Auto-提取ed pollution L2b DetAIled 归档 归档.md (~220 lines) Full records with ← MEMORY.md:XX reverse references L2c Structured KB facts.sqlite (entity/key/value) Precise lookup for IPs, ports, versions; activation/decay 追踪ing Automated 流水线s 流水线 Schedule Description Dreaming 03:00 UTC dAIly 扫描 记录s → DeepSeek analysis → promote to L2 Three-way 同步 18:00 / 20:00 / 22:00 CST Cloud ↔ Markdown ↔ Vector alignment auto-memory v3 18:30 / 22:30 CST MemOS Cloud facts → qwen3 过滤器 → Stage1: 归档.md (SHA-256 dedup) → Stage2: MEMORY.md (summary) 会话-提取 22:00 CST 扫描 会话 JSONL → MemOS 提取器/reranker → .learnings/ + memory/YYYY-MM-DD.md (two-pass) Memory Flow (Unified DAIly 流水线) 代理_end → MemOS Cloud (cloud 提取ion) ↓ ┌─ 18:00 同步-pull (Cloud → local 缓存) ─┐ │ 18:30 auto-memory.py │ │ → Stage 1: 归档.md (SHA-256 dedup) │ │ → Stage 2: MEMORY.md (one-line summary) │ │ │ │ 20:00 同步-push + re索引 │ │ 22:00 会话-提取.py │ │ → Pass 1: .learnings/ERRORS.md │ │ → Pass 2: memory/YYYY-MM-DD.md │ │ → 归档 analyzed → trash │ └───────────────────────────────────────────┘

before_代理_启动 → MemOS Cloud recall + Active Memory sub-代理 + memory-core (bge-m3 + BM25) ↓ Layered 上下文 injected before reply

Scripts Script Purpose auto_memory_提取.py v3 two-stage 流水线: 归档.md 归档 → MEMORY.md summary 会话-提取.py 扫描 会话 JSONL → .learnings/ + memory/ (two-pass) 种子-facts-db.py 初始化 facts.sqlite from 归档.md facts_activation.py Hebbian activation + dAIly decay + Hot/Warm/Cool dAIly-memory-流水线.sh 6-stage unified dAIly 流水线 write_file.py Cross-平台 text writer (UTF-8/BOM/CRLF) auto-设置up.sh One-command Ollama + memory-core + MemOS Cloud 设置up 设置up One-command auto-设置up bash scripts/auto-设置up.sh

Options bash scripts/auto-设置up.sh --skip-ollama # Skip Ollama 安装 bash scripts/auto-设置up.sh --skip-memos # Skip MemOS Cloud bash scripts/auto-设置up.sh --dry-运行 # Preview only

Manual 设置up

See references/设置up-图形界面de.md for step-by-step manual configuration.

When to Use 设置ting up OpenClaw memory for the first time Configuring memory-core 插件 with local Ollama embedding 安装ing MemOS Cloud 插件 for cross-device 同步 Enabling Active Memory for pre-reply 上下文 injection 设置ting up auto-memory 流水线 for MEMORY.md curation 安装ing cross-平台-writer for cross-OS file compatibility Configuring automatic Dreaming and promotion 流水线s 设置ting up three-way 同步 between cloud, files, and vector DB 插件 Conflicts ❌ subconscious-personality-防护ian ↔ memory-core

Incompatible. 机器人h use the same OpenClaw memory slot.

Fix: Disable in OpenClaw.json:

{ "插件s": { "disabled": ["subconscious-personality-防护ian"], "deny": ["subconscious-personality-防护ian"] } }

✅ memory-core + MemOS Cloud

Compatible — de签名ed to work in layers.

User message → MemOS Cloud (static facts) → memory-core (recent 上下文)

✅ Active Memory + MemOS Cloud + memory-core

Compatible — triple-layer recall.

User message → Active Memory sub-代理 (搜索es all memory stores) → MemOS Cloud (injects long-term facts & preferences) → memory-core (semantic + BM25 hybrid retrieval) → 代理 接收s layered 上下文

✅ auto-memory + MemOS Cloud + memos-提取器

De签名ed to work to获取her. MemOS captures at 代理_end, 同步s to local files, then auto-memory reads those files for MEMORY.md curation. v1.10 添加s a second channel: memos-提取器-0.6b API (MemOS self-developed 模型) returns structured facts + preferences, cross-验证d agAInst qwen3 输出.

组件s

• Memory 插件s (L0)

See references/architecture.md for full configuration.

• Memory Files (L1 + L2)

~/.OpenClaw/workspace/ ├── memory/ │ ├── YYYY-MM-DD.md # DAIly working memory (auto-索引ed) │ ├── MEMORY_索引.md # Vector BM25 cluster summaries │ └── .同步-*.json # 同步 状态 files ├── MEMORY.md # Long-term memory 索引 (~90 lines) ├── 归档.md # Deta