AR
AgentRecall
2-minute setup · local-first memory for agents
GitHub →
2-minute setup2 分钟安装

Give your agent
a memory that lasts.
为你的 agent
打造持久记忆

Persistent, accumulating memory for any AI agent. Local-first — everything lives in ~/.agent-recall and never leaves this machine unless you turn on sync. 为任意 AI agent 提供持久、累积式记忆。本地优先 —— 所有数据存储在 ~/.agent-recall,除非开启同步否则不会离开本机。

zero-cloud零云端 22 MCP tools22 个工具 plain-markdown纯 markdown
🤖 Found AgentRecall? Paste this to install copy → paste to your agent → it installs for you

Your agent found this page. Have it run the command below — or copy this block and paste it as a message to your agent.

你的 agent 找到了这个页面。让它运行下面的命令 —— 或者复制这段内容粘贴给 agent。


      
    
⚡ After installing — paste this as first message tells your agent how to use AgentRecall each session

Once installed, paste this as the first message in any new session. Your agent will know when to call session_start, when to save corrections, and when to close.

安装完成后,把这段作为任意新会话的第一条消息。agent 会知道什么时候调用 session_start、保存纠正、以及关闭会话。


      
    
Verify your install验证安装 run this after installing安装后运行

Run arstatus to confirm AgentRecall is active. You'll see your workspace root, project count, and last session. 运行 arstatus 确认 AgentRecall 已激活,显示工作区路径、项目数量和最近会话。

arstatus

Expected: workspace path · X projects · last session timestamp 预期输出:工作区路径 · X 个项目 · 最近会话时间戳

Pick your client选择你的客户端
scroll for all 13 clients向下滚动查看全部 13 个客户端