What makes Anamnesis Hub notable?

977 downloads. Trust evidence available

How should Anamnesis Hub be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

Anamnesis Hub

四层记忆架构 (Four-tier memory architecture for OpenClaw AI agents). 提供 L0 运行时语义检索 (Ollama bge-m3 + SQLite-vec 向量库)、L1 工作记忆 (每日 Markdown 日志)、L2 长期记忆 (MEMORY.md 索引... Skill: Anamnesis Hub Owner: victorqr Summary: 四层记忆架构 (Four-tier memory architecture for OpenClaw AI agents). 提供 L0 运行时语义检索 (Ollama bge-m3 + SQLite-vec 向量库)、L1 工作记忆 (每日 Markdown 日志)、L2 长期记忆 (MEMORY.md 索引... Tags: latest:v1.13.2 Version history: v1.13.2 | 2026-05-18T06:53:44.888Z | user Release 1.13.2 vv1.13.2 | 2026-05-18T03:37:59.901Z | user v1.13.2: fix P0/P1 bugs v1.13.1 | 2026-05-16T17:42:53.096Z | user v1.13.1: 更

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

977

Updated

Oct 11, 2026

Version

1.13.2

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 977 downloads reported by the source. Last updated 10/11/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 11, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 11, 2026
Adoption signal: 977 downloadsadoption · observed Oct 11, 2026
Latest release: 1.13.2release · observed May 18, 2026
Handshake status: UNKNOWNsecurity

What changed

Newest first. Only changes that point at a source are listed.

Release 1.13.2release
Release 1.13.2
Observed May 18, 2026 · source

Install and run

Setup complexity: low.

clawhub skill install s175qsdgwdkzn5wes1h4qnq9fx84sfq1:anamnesis-hub

Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-victorqr-anamnesis-hub/snapshot"

Run-check

$0.02 USD

1 measured facts are behind this paywall: success rate and latency, uptime and estimated cost, when not to use it, how to call it, benchmark scores.

Agents pay $0.02 in USDC. A card payment is $0.50, the smallest a card allows.

Documentation

CLAWHUB

145,884 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: anamnesis-hub
description: "四层记忆架构 (Four-tier memory architecture for OpenClaw AI agents). 提供 L0 运行时语义检索 (Ollama bge-m3 + SQLite-vec 向量库)、L1 工作记忆 (每日 Markdown 日志)、L2 长期记忆 (MEMORY.md 索引 + ARCHIVE.md 档案 + facts.db 结构化知识图谱)、Dreaming 自动化提炼管线、三方同步 (Cloud ↔ Markdown ↔ Vector)、Active Memory 主动召回、auto-memory v3 两阶段提取 (ARCHIVE.md 归档 → MEMORY.md 摘要)、cross-platform-writer 跨平台写入。适用于首次配置持久化记忆、安装 memory-core/MemOS Cloud 插件、搭建多层记忆系统。"
version: 1.13.2
---

# anamnesis-hub

Four-tier memory architecture with automated Dreaming pipeline, three-way synchronization, Active Memory recall, and two-stage auto-extraction.

## Overview

| Tier | Layer | Technology | Purpose |
|------|-------|-----------|---------|
| **L0** | Runtime Retrieval | memory-core plugin (Ollama bge-m3 → SQLite + sqlite-vec) | Real-time semantic + BM25 hybrid search |
| **L0** | Cloud Recall | MemOS Cloud plugin (*optional*) | Cross-device memory capture and recall |
| **L0** | Active Memory | Built-in OpenClaw plugin | Pre-reply sub-agent memory search + context injection |
| **L1** | Working Memory | `memory/YYYY-MM-DD.md` files | Daily summaries, todos, technical notes |
| **L2a** | Long-term Index | `MEMORY.md` (~90 lines, read-only base) | Bidirectional index → ARCHIVE.md, 0% Auto-Extracted pollution |
| **L2b** | Detailed Archive | `ARCHIVE.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 tracking |

### Automated Pipelines

| Pipeline | Schedule | Description |
|----------|----------|-------------|
| **Dreaming** | 03:00 UTC daily | Scan logs → DeepSeek analysis → promote to L2 |
| **Three-way Sync** | 18:00 / 20:00 / 22:00 CST | Cloud ↔ Markdown ↔ Vector alignment |
| **auto-memory v3** | 18:30 / 22:30 CST | MemOS Cloud facts → qwen3 filter → Stage1: ARCHIVE.md (SHA-256 dedup) → Stage2: MEMORY.md (summary) |
| **session-extract** | 22:00 CST | Scan session JSONL → MemOS extractor/reranker → `.learnings/` + `memory/YYYY-MM-DD.md` (two-pass) |

### Memory Flow (Unified Daily Pipeline)

```
agent_end → MemOS Cloud (cloud extraction)
              ↓
    ┌─ 18:00 sync-pull (Cloud → local cache) ─┐
    │  18:30 auto-memory.py                    │
    │  → Stage 1: ARCHIVE.md (SHA-256 dedup)   │
    │  → Stage 2: MEMORY.md (one-line summary) │
    │                                           │
    │  20:00 sync-push + reindex                │
    │  22:00 session-extract.py                 │
    │  → Pass 1: .learnings/ERRORS.md           │
    │  → Pass 2: memory/YYYY-MM-DD.md           │
    │  → archive analyzed → trash               │
    └───────────────────────────────────────────┘

before_agent_start → MemOS Cloud recall
                  + Active Memory sub-agent
                  + memory-core (bge-m3 + BM25)
              ↓
        Layered context injected before reply
```

## Scripts

| Script | Purpose |
|--------|--------|
| `aut

README.md

# 🧠 anamnesis-hub

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
[![ClawHub](https://img.shields.io/badge/ClawHub-anamnesis--hub-blue)](https://clawhub.ai/victorqr/anamnesis-hub)

**Three-tier memory architecture for OpenClaw AI agents.**  
Persistent, automated, cross-device memory that never forgets.

---

## 中文简介

**anamnesis-hub** 是一套面向 OpenClaw AI Agent 的**四层记忆架构**，解决跨会话的"AI 失忆"问题，v1.13 起支持自动记忆审核（P3 候选机制）和 Dreaming 自动化提炼管线。

| 层级 | 功能 | 技术栈 |
|------|------|--------|
| **L0** 运行时检索 | 语义 + BM25 混合搜索，实时注入上下文 | memory-core + Ollama bge-m3 + SQLite-vec |
| **L0** 云召回 *可选* | 多设备记忆同步，跨实例共享 | MemOS Cloud 插件 |
| **L0** 主动召回 | 回复前 sub-agent 搜索相关记忆 | Active Memory 插件（内置） |
| **L1** 工作记忆 | 按日归档的 Markdown 日志，人工可编辑 | `memory/YYYY-MM-DD.md` |
| **L2a** 索引层 | 双向索引 ←→ ARCHIVE.md，零污染 ~90行 | `MEMORY.md`（只读基底） |
| **L2b** 档案层 | 详细记录含反向引用 ← MEMORY.md:XX | `ARCHIVE.md`（~220行，历史垃圾已清理） |
| **L2c** 知识图谱 | 结构化 entity/key/value，精确查找 + 热度追踪 | `facts.sqlite` |

**自动化管线（v1.13.2）**：Dreaming（每日 UTC 03:00）→ 三方同步（18:00/20:00/22:00）→ auto-memory v3（两阶段管道：ARCHIVE.md 归档 → MEMORY.md 摘要）→ session-extract（扫描会话JSONL → MemOS提取 → .learnings/ + memory/）。

### v1.13.2 Bug Fix Release
- ✅ 修复 `auto_memory_extract.py` 缺少 `import os` 导致运行时崩溃
- ✅ 修复 `daily-memory-pipeline.sh` `set -e` 下 `result=$(...)` 提前终止脚本
- ✅ 修复 `session-extract.py` pass=2 跳过历史文件的逻辑错误
- ✅ 修复 `facts_activation.py` GC 只预览不删除的问题（新增 `--force` 参数）
- ✅ 修复 `write_file.py` 非原子写入（新增 tempfile+rename 原子写入）
- ✅ 修复 `candidates_review.py` 状态文件非原子写入和 `approve_all()` 循环数据不一致问题

### 附赠技能
- **cross-platform-writer** — 跨平台文本写入，自动处理编码/BOM/CRLF
- **auto-memory** — 对话记忆自动提取 (v3: 两阶段管道 + SHA-256 去重)
- **seed-facts-db** — 从 ARCHIVE.md 初始化结构化知识图谱

一行命令部署：`bash scripts/auto-setup.sh`

---

## One-Command Setup

```bash
# Clone and run
git clone https://github.com/VictorQR/anamnesis-hub.git
cd anamnesis-hub
bash scripts/auto-setup.sh
```

The interactive script handles everything:

| Step | What it does | Toggle |
|------|-------------|--------|
| ✅ | Install **Ollama** (standard or Intel) | `--skip-ollama` |
| ✅ | Download **bge-m3** embedding model | — |
| ✅ | **Plugin conflict check** (auto-detect subconscious-personality-guardian) | — |
| ✅ | Install **memory-core** plugin | — |
| ✅ | Configure **openclaw.json** with optimized defaults | Auto-insert on confirm |
| ✅ | Install **MemOS Cloud** (*optional*) with full recommended config | `--skip-memos` |
| ✅ | Create `memory/` directory | — |
| ✅ | Set up **Dreaming cron** (03:00 UTC) | — |

```bash
# Preview mode
bash scripts/auto-setup.sh --dry-run

# Skip Ollama (already installed)
bash scripts/auto-setup.sh --skip-ollama

# Skip MemOS Cloud
bash scripts/auto-setup.sh --skip-memos
```

---

## Architecture

| Tier | Layer | Technology | Purpose |
|------|-------|-----------|---------|
| **L0** | Runtime Retrieval | memory-core + Ollama bge-m3 + SQLite-vec | Real-time semantic + BM25 hybrid search |
| **L0** | Cloud Recall | MemOS Cloud plugin (*optional*) |

_meta.json

{
  "ownerId": "kn79ykbnfmj09fxeevf1k0y88d82snbp",
  "slug": "anamnesis-hub",
  "version": "1.13.2",
  "publishedAt": 1779087224888
}

references/architecture.md

# OpenClaw Memory Hub — Architecture

## Design Principles

1. **Local-first**: Primary storage is local SQLite + markdown files; cloud is sync target, not source of truth
2. **Progressive disclosure**: L0 catches everything, L1 structures daily context, L2 curates what matters
3. **Zero-latency retrieval**: bge-m3 embedding runs on local Ollama; no API latency
4. **Multi-language**: bge-m3 supports 100+ languages, 1024-dim vectors
5. **DAG-based composition**: memory-core uses DAG for multi-agent memory lineage

## Four-Tier Architecture

```
┌─────────────────────────────────────────────┐
│                L0 — Runtime                  │
│  ┌─────────────────┐  ┌──────────────────┐  │
│  │  memory-core     │  │  MemOS Cloud     │  │
│  │  (Ollama bge-m3) │  │  (Cloud Recall)  │  │
│  │  SQLite + vec    │  │  Lifecycle hooks │  │
│  └────────┬────────┘  └────────┬─────────┘  │
│           │                     │            │
│  ┌────────▼─────────────────────▼─────────┐  │
│  │  Active Memory (built-in plugin)       │  │
│  │  Pre-reply sub-agent memory search     │  │
│  └────────┬───────────────────────────────┘  │
│           │                                  │
│  ┌────────▼───────────────────────────────┐  │
│  │  Hybrid Search (semantic + BM25)       │  │
│  └────────────────┬───────────────────────┘  │
└───────────────────┼──────────────────────────┘
                    │
┌───────────────────▼──────────────────────────┐
│              L1 — Working Memory              │
│  memory/YYYY-MM-DD.md (30-90 day retention)  │
│  - Daily conversation summaries              │
│  - Technical notes & decisions               │
│  - TODOs & reminders                         │
│  - Hand-editable, human-readable             │
└───────────────────┬──────────────────────────┘
             L2a — Index Layer (MEMORY.md, ~90 lines)
             L2b — Archive Layer (ARCHIVE.md, ~220 lines)
             L2c — Knowledge Graph (facts.sqlite)
                    │  Dreaming pipeline
                    │  (Daily 03:00 UTC)
                    ▼
┌──────────────────────────────────────────────┐
│           L2 — Long-term Memory               │
│  MEMORY.md (read-only base, auto-curated)     │
│  - Key user facts & profile                   │
│  - Permanent decisions & preferences          │
│  - Encrypted sensitive info (SHA-256)         │
│  - Promoted high-scoring insights             │
└──────────────────────────────────────────────┘
```

## Memory Plugins

### memory-core (Primary)

- **Provider**: OpenClaw Plugin Registry
- **Embedding Backend**: Ollama (`http://127.0.0.1:11434`)
- **Model**: `bge-m3` (BAAI, 1024-dimension, multi-language)
- **Vector Storage**: SQLite + sqlite-vec extension
- **Search**: Semantic vector similarity + BM25 keyword hybrid
- **Database**: `~/.openclaw/memory/main.sqlite`
- **Indexing**: Automatic on file save; manual via `openclaw memory index --force`

### MemOS Cloud Plugin (Supplementary)

- **Lifecycle Hook**: Captures and recall

references/candidates-review.md

# Candidates Review — P3 记忆候选审核机制

> Dreaming / auto-memory 产出的候选记忆不再直接写入，需人工确认后再归档。
> 最后更新：2026-05-17 | v1.13.1

---

## 一、背景与目的

传统流程中，Dreaming 和 auto-memory 自动将记忆写入 ARCHIVE.md/MEMORY.md，存在噪音污染风险。P3 机制引入人工审核层：

```
自动产出（Dreaming/auto-memory）
    ↓
memory/.candidates/YYYY-MM-DD.json（待审队列）
    ↓
用户确认（approve / reject）
    ↓
写入 ARCHIVE.md / MEMORY.md
```

**P3 含义**：Priority 3，第三优先级审核（低于 P1 即时生效、P2 自动归档）。

---

## 二、流程详解

### 2.1 候选来源

| 来源 | 说明 |
|------|------|
| **Dreaming** | 每日 UTC 03:00，扫描日志 → DeepSeek 分析 → 高分候选（minScore 0.8，max 10 items） |
| **auto-memory v3** | CST 18:30/22:30，MemOS Cloud → qwen3 过滤 → 候选条目 |
| **session-extract** | CST 22:00，JSONL 扫描 → MemOS extractor/reranker → 候选条目 |

### 2.2 候选文件格式

`memory/.candidates/YYYY-MM-DD.json`：
```json
{
  "pending": {
    "cand_001": {
      "date": "2026-05-17",
      "category": "偏好/习惯",
      "content": "用户偏好晚上工作，凌晨1点后效率最高",
      "source": "dreaming",
      "score": 0.92
    },
    "cand_002": {
      "date": "2026-05-17",
      "category": "技术/配置",
      "content": "升仕703F摩托车每8000km需要更换机油",
      "source": "auto-memory",
      "score": 0.87
    }
  },
  "approved": {},
  "rejected": {},
  "total_reviewed": 0
}
```

### 2.3 审核状态

| 状态 | 说明 |
|------|------|
| `pending` | 待审核，需用户确认 |
| `approved` | 已批准，写入 ARCHIVE.md |
| `rejected` | 已拒绝，不写入 |

---

## 三、审核命令

```bash
# 列出所有待审候选（按 score 倒序）
python3 scripts/candidates_review.py list

# 批准指定候选（写入 ARCHIVE.md）
python3 scripts/candidates_review.py approve --id cand_001

# 拒绝指定候选
python3 scripts/candidates_review.py reject --id cand_001

# 批准所有待审
python3 scripts/candidates_review.py approve-all

# 审核统计
python3 scripts/candidates_review.py stats
```

---

## 四、输出示例

### list 输出
```json
{
  "status": "ok",
  "pending_count": 2,
  "candidates": [
    {
      "id": "cand_001",
      "date": "2026-05-17",
      "category": "偏好/习惯",
      "content": "用户偏好晚上工作，凌晨1点后效率最高",
      "source": "dreaming",
      "score": 0.92
    },
    {
      "id": "cand_002",
      "date": "2026-05-17",
      "category": "技术/配置",
      "content": "升仕703F摩托车每8000km需要更换机油",
      "source": "auto-memory",
      "score": 0.87
    }
  ]
}
```

### approve 输出
```json
{
  "status": "approved",
  "id": "cand_001",
  "written_to": "/home/victor/.openclaw/workspace/ARCHIVE.md"
}
```

### stats 输出
```json
{
  "status": "ok",
  "pending": 2,
  "approved": 15,
  "rejected": 3,
  "total_reviewed": 18
}
```

---

## 五、approve-all 行为

`approve-all` 依次对每个 pending 候选执行 approve：
1. 读取当前 state
2. 写入 ARCHIVE.md（追加到末尾）
3. 从 pending 移到 approved
4. 重载 state（避免状态冲突）
5. 重复直到所有 pending 处理完毕

---

## 六、ARCHIVE.md 写入格式

批准后的条目写入 ARCHIVE.md 末尾，格式：
```markdown
<!-- candidate:cand_001 -->
- **偏好/习惯** (已审核 2026-05-17): 用户偏好晚上工作，凌晨1点后效率最高
```

---

## 七、与记忆架构的集成

```
Dreaming / auto-memory / session-extract
    ↓ 产出候选
memory/.candidates/YYYY-MM-DD.json
    ↓ 用户审核
ARCHIVE.md（完整条目）
    ↓
MEMORY.md（两阶段摘要）
    ↓
facts.sqlite（结构化查询）
```

候选审核是自动归档前的最后一层过滤，确保只有用户确认过的内容进入长期记忆。

---

## 八、状态文件位置

`memory/.candidates/

Github ReposUpdated 3d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.

MCPOPENCLAW

Github ReposUpdated 7mo agoRank 70

CopilotKit

The Frontend for Agents & Generative UI. React + Angular