persistent-memory

@rrrrrredy⭐ 0 stars

Use when the owner asks the agent to remember or recall durable personal/work facts ("帮我记住", "记一下", "别忘了", "你还记得吗"), initialize/check the memory system, or distill daily logs into facts.yaml and MEMORY.md. Trigger for memory persistence, facts extraction, memory hygiene, daily log distillation, and agent recall workflows. Not for group or non-owner sessions, one-off Q&A, vector database setup, or unrelated note-taking systems.

—/10

// RATINGS

⭐GitHub Stars

⭐ 0GitHub ↗

New / niche

🟢ProSkills Score

—

📍

Not yet listed on ClawHub or SkillsMP

// README

persistent-memory

Agent persistent memory system with 3-layer distillation: daily logs → facts.yaml → MEMORY.md.

OpenClaw Skill — works with OpenClaw AI agents

What It Does

Gives your OpenClaw agent long-term memory that survives across sessions. Uses a 3-layer distillation architecture: raw daily conversation logs are automatically summarized into structured daily notes, atomic facts are extracted into a YAML knowledge base, and key insights are periodically refined into a compact MEMORY.md (≤80 lines) loaded every session. Includes automated cron jobs for daily digestion and weekly distillation, plus on-demand "remember this" and "do you remember" workflows.

Quick Start

openclaw skill install persistent-memory
# Or:
git clone https://github.com/rrrrrredy/persistent-memory.git ~/.openclaw/skills/persistent-memory

Then initialize:

python3 ~/.openclaw/skills/persistent-memory/scripts/memory_manager.py init

Features

3-layer memory architecture: Daily logs → facts.yaml (atomic facts) → MEMORY.md (core cognition)
Automated daily digestion: 00:00 cron reads session JSONL, generates structured summaries
Weekly distillation: Friday 18:00 cron extracts new facts, prunes MEMORY.md, archives old logs
On-demand memory: "帮我记住 X" instantly stores facts or events in the right layer
Semantic search: "你还记得 X 吗" searches across all memory layers with memory_search()
Health checks: Validates file structure, MEMORY.md line count, facts.yaml format
Auto-archiving: Logs older than 30 days are moved to memory/archive/
Safety-first: Memory operations blocked in group chats; deletions require user confirmation

Usage

Trigger phrases:

帮我记住, 记一下, 别忘了 — store information
你还记得, 上次说的 — recall information
初始化记忆系统, 设置记忆 — first-time setup
检查记忆, 记忆系统健康吗 — health check

Project Structure

persistent-memory/
├── SKILL.md                    # Main skill documentation (430 lines)
├── MEMORY.md.template          # Template for MEMORY.md initialization
├── USER.md.template            # Template for USER.md initialization
├── scripts/
│   └── memory_manager.py       # CLI tool: init, digest, write-daily, facts, health, archive
└── references/
    └── cron-setup.md           # Cron configuration guide & facts.yaml format template