Token导航 LogoToken导航TokenDH.com
研究检索执行命令clawhub未标认证来源可访问clear审计通过

wal-memory沃尔记忆

Agent Skill

wal-memory 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

9,573

周安装

407

GitHub Stars

公开资料未说明

下载量

3,354
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:wal-memory(沃尔记忆)
来源仓库:https://github.com/maikunari/wal-memory
安装命令:
openclaw skills install wal-memory
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install wal-memory

简介

wal-memory 使用预写日志系统实现会话崩溃恢复和内存压缩。

  • 适合在 OpenClaw 中为代理配置持久化内存和故障恢复机制时使用。
  • 支持两文件 WAL 结构,确保数据一致性和快速重启。
  • 安装前需确认权限范围、维护状态及是否会触发磁盘写入操作。
  • 适用于需要高可靠性会话管理的开发场景。wal-memory 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

name
wal-memory
description
Session crash and compaction recovery using a two-file WAL (Write-Ahead Log) system. Use when setting up persistent memory for an OpenClaw agent that needs to survive session rollovers, gateway disconnects, Anthropic outages, or context compaction. Installs state-log.js + GOALS.md structure + cold boot recovery hook. Eliminates agent amnesia between sessions.
metadata
requires
Node.js (node binary must be available in PATH)

WAL Memory

Two-file system for session crash recovery. Survives gateway disconnects, provider outages, and context compaction. Co-designed with Gemini.

The Problem

OpenClaw agents lose state in two ways:

  1. Compaction — context window fills, gets summarized, detail lost
  2. Session rollover — session dies (crash/disconnect/overload), new session starts cold with zero history

The Two Files

  • memory/GOALS.md — The "why": current milestone, sprint tasks, architecture decisions. Updated manually at major milestones or end of session.
  • STATE.log — The "what just happened": append-only WAL, auto-timestamped. Never edit manually. Written via state-log.js.

Setup

1. Install the logger script

Copy scripts/state-log.js to ~/clawd/scripts/state-log.js.

Test it:

node ~/clawd/scripts/state-log.js "STARTUP" "WAL memory system initialized"
# Should print: Logged: STARTUP
# STATE.log should appear in ~/clawd/

2. Create GOALS.md

Copy assets/GOALS.template.md to ~/clawd/memory/GOALS.md. Fill in your current objective, sprint tasks, and key decisions. Keep it under 30 lines — this is the "why", not the "what".

3. Update AGENTS.md — Cold Boot Recovery Hook

Add this to your "Every Session" startup sequence:

## Every Session — Cold Boot Recovery Hook
1. Read `memory/GOALS.md` — current mission and what's in flight
2. Run `tail -n 20 ~/clawd/STATE.log` — last 20 actions, tells you where we left off
3. Read today's `memory/YYYY-MM-DD.md` for recent context

4. Update HEARTBEAT.md — Safety Net Flush

Add this as the first step of every heartbeat:

node ~/clawd/scripts/state-log.js "HEARTBEAT" "Interval reached $(date '+%Y-%m-%d %H:%M')"

This guarantees STATE.log has a timestamp even if the session crashes immediately after — you'll always know exactly when things flatlined.

The Logging Rule

After any meaningful action, run:

node ~/clawd/scripts/state-log.js "CATEGORY" "What you just did"

Categories:

  • ACTION — file created, code edited, config changed
  • DEPLOY — pushed to production, service restarted
  • AGENT — dispatched or agent task completed
  • TASK — milestone completed
  • FIX — bug or system fix applied
  • STARTUP — cold boot (always log this first)
  • HEARTBEAT — interval ping
  • ERROR — something failed

Examples:

node ~/clawd/scripts/state-log.js "AGENT" "Dispatched Owsley to build auth system"
node ~/clawd/scripts/state-log.js "DEPLOY" "FF ElasticSearch secured, 6027 docs indexed"
node ~/clawd/scripts/state-log.js "FIX" "Cron syntax fixed — was using session spawn, now uses agent --deliver"

Do NOT log: reads, searches, minor tool calls. Signal only. Never log: passwords, API keys, tokens, or sensitive payloads — STATE.log is plaintext.

Recovery Protocol

On cold boot or post-compaction, first action is always:

# 1. Read the why
cat ~/clawd/memory/GOALS.md

# 2. Read what just happened
tail -n 20 ~/clawd/STATE.log

# 3. Resume from last log entry

File Details

  • STATE.log lives at ~/clawd/STATE.log
  • Auto-rotates at 5MB (renames to STATE.log.old, starts fresh)
  • Format: [2026-03-03T09:00:00.000Z] [CATEGORY] Message
  • Add STATE.log to .gitignore — it may contain sensitive action descriptions, credentials referenced in log messages, or private task details. Do not commit it to source control unless you have reviewed its contents and are certain it contains no sensitive data.

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

93.42%
按下载量换算3,133

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 openclaw skills install wal-memory 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills