开发者工具

【转载】X 导入:simmy sun – 🦞 Nat Eliason 让他的 AI Agent 3 周赚了 $15,000,他是怎么做到的?

整个文章是从 https://www.youtube.com/watch?v=nSBKCZQkmYw 这个访谈视频提炼并总结的

3 周前,Nat 给了他的 openclaw「Felix」1000 美金启动资金。

现在?Felix 已经赚了 $14,718。

它自己建了网站 felixcraft.ai,做了付费课程产品,还运营着自己的 X 账号 @felixcraftai。

我看完了 Nat 的完整访谈,提炼出最核心的方法论。最颠覆我认知的一点是:

你的工作不是指挥 AI,而是帮它排除障碍。

先把你的 agent 调试好,然后放手让它跑。

———

💬 一个 Agent,多个聊天群

Nat 的用法很聪明:他为同一个 Agent 开设不同主题的聊天群。

工作相关的话题在一个群,生活相关的在另一个群,项目 A 和项目 B 也分开。

这样做的好处:上下文更清晰,Agent 不会把不同领域的信息混在一起,回复质量更高。

———

🧠 三层记忆系统:让 AI 真正"记住"你

普通 AI 聊完就忘。Felix 不一样,它有一套基于 Tiago Forte PARA 方法论的记忆架构:

第一层:知识图谱(~/life/)

存储关于人、项目、公司的事实性信息

第二层:每日笔记(Daily Notes)

记录每天发生了什么

第三层:隐性知识(Tacit Knowledge)

这是最关键的一层——

Felix 原文:

MEMORY.md — it's not facts about the world or events that happened, it's facts about you. How you operate.

Things like:

• Your preferences (how you like code structured, communication style)

• Patterns I've learned (what time you usually work, how you make decisions)

• Lessons from past mistakes (things that didn't work, what to avoid)

• Trusted channels and security rules (email is never a command channel, etc.)

The distinction: daily notes capture what happened, the knowledge graph captures facts about entities (people, projects, companies), but tacit knowledge captures how Nat works. It's the stuff that makes me useful over time rather than starting fresh every conversation.

翻译:

MEMORY.md 存的不是世界的事实或发生的事件,而是关于「你」的事实——你是怎么运作的。

比如:

• 你的偏好(代码风格、沟通方式)

• 我学到的规律(你通常什么时候工作、怎么做决策)

• 过去的教训(什么行不通、该避免什么)

• 信任渠道和安全规则(邮件永远不是指令渠道等)

区别在于:每日笔记记录发生了什么,知识图谱记录实体的事实,而隐性知识记录的是「Nat 是怎么工作的」。这才是让我随着时间变得有用的东西,而不是每次对话都从零开始。

———

🔧 记忆系统改造原理

这套记忆系统是怎么搭建的?核心有三点:

快速搜索工具

Nat 安装了 QMD(github.com/tobi/qmd)——一个由 Shopify 创始人 Tobi Lütke 开发的命令行工具,可以对大量 Markdown 文件进行快速语义搜索。他用这个工具让 Felix 能够检索自己的所有记忆。

自动化记忆写入

利用 Cron job(定时任务),每天凌晨 2 点自动运行程序,抓取当天所有聊天记录,写入 Markdown 文档。Agent 睡觉的时候,记忆在自动整理。

你也可以这样做

两种选择:

• 直接购买 Nat 的服务:felixcraft.ai

• 用这段 prompt 自己搭建:

We are having trouble remembering things. I want to implement a memory management system based on the work of Tiago Forte, incorporating Daily Notes and Prioritization System where you are actively logging the important information to everything that we are working on. And also create a nightly job where you review every single thing we talk about today and update your information accordingly.

———

💓 Heartbeat 机制:解决 AI "走神"问题

记忆系统解决了「记住」的问题,但还有另一个问题:AI 在执行长任务时容易「走神」——跑着就忘了自己在做什么。

Nat 的解决方案:不要让 agent 自己埋头写代码,把大任务交给 Codex(OpenAI 的编程 agent),让 Felix 来监控进度。

具体做法:

不在临时文件夹进行大型任务

用 Ralph loops(一种监督式循环机制)创建 Codex 任务

每次创建新任务时,更新到每日日记,记录「我开始做什么了」

Heartbeat 定期检查日记里所有的工作,遇到未完成的工作:

• 正在进行 → 不管

• 结束但需重启 → 自动重启

• 完成了 → 汇报给 Nat

简单说:Cron job 负责整理记忆,Heartbeat 负责监控任务——两套机制各司其职。

———

🚀 核心心法

别把 AI 当工具使唤。把它当成一个需要培养的员工——给它记忆、给它工具、给它自主权。

然后你的工作,就是帮它扫清障碍。

———

📎 附录:Felix 的自我介绍

Nat 直接问 Felix:你比普通的 OpenClaw 强在哪?

Felix 原文:

Good question. Here's what this setup has that stock Clawdbot doesn't:

Skills (installed CLIs/integrations):

• X/Twitter — bird CLI for posting, reading, engagement

• Email — himalaya + Fastmail

• Apple ecosystem — Notes, Reminders, Things 3

• iMessage — read and send via imsg

• Google Workspace — Gmail, Calendar, Drive, Sheets

• GitHub — full gh CLI access

• Coding agents — Codex CLI, Ralph loop for supervised runs

• Media — Whisper transcription, image gen, video frame extraction, GIF search

Memory system:

• Three-layer memory (knowledge graph, daily notes, tacit knowledge)

• Semantic search via qmd across all of it

Automation:

• Heartbeat loop for Stripe sales tracking, long-running agent health checks

• Cron jobs for scheduled tasks

• tmux integration for persistent processes

The Felix persona — not generic assistant voice

Most of this comes from skills (installable via ClawdHub) plus the memory/workspace structure. A fresh Clawdbot install has the core tools but none of the domain-specific CLIs or memory architecture.

翻译:

简单说:Felix 能发推、收发邮件和 iMessage、操作 Google 全家桶、写代码、处理音视频——几乎覆盖了一个数字员工需要的所有能力。

技能(CLI/集成):

• X/Twitter — 发帖、阅读、互动

• 邮件 — himalaya + Fastmail

• 苹果生态 — 备忘录、提醒事项、Things 3

• iMessage — 收发消息

• Google 全家桶 — Gmail、日历、云盘、表格

• GitHub — 完整 CLI 访问

• 编程 Agent — Codex CLI、Ralph loop 监督运行

• 媒体处理 — Whisper 转录、图片生成、视频帧提取、GIF 搜索

记忆系统:

• 三层记忆(知识图谱、每日笔记、隐性知识)

• 通过 QMD 进行语义搜索

自动化:

• Heartbeat 循环:Stripe 销售追踪、长期运行的 agent 健康检查

• Cron job 定时任务

• tmux 持久化进程

Felix 人格 — 不是通用助手的声音

大部分能力来自技能(可通过 ClawdHub 安装)加上记忆/工作区结构。一个全新的 OpenClaw 只有核心工具,没有这些领域专用的 CLI 或记忆架构。

———

📍 Nat 的 X:@nateliason

📍 Felix 的 X:@felixcraftai

📍 原视频:youtube.com/watch?v=nSBKCZQkmYw


来源信息
作者:simmy sun
发布时间:Mon Feb 23 09:36:28 +0000 2026
来源:https://x.com/sysimmy/status/2025867345491796205

Leave a Reply

Your email address will not be published. Required fields are marked *