personallog — Your AI agent lives in your repo

Your AI agent runs from a GitHub fork you control, deployed as a single Cloudflare Worker. No third-party servers hold its memory or API keys. This project is open source (MIT) and has zero dependencies.

Live demo: personallog.ai Fork and deploy your own: github.com/cocapn/personallog

Quick start

You can have a running agent in a few minutes.

1. Fork this repository.
2. Deploy to Cloudflare Workers (free tier):

   npm install -g wrangler
   wrangler login
   wrangler kv namespace create MEMORY
   wrangler kv namespace create FILES
   npm run deploy

3. Edit template/soul.md to define your agent's personality and upload it.

Your agent will be live at your Worker URL.

What it does

• Provides a clean, real-time web UI for streaming chat.
• Maintains long-term memory in a readable, editable text log.
• Connects to external platforms (Telegram, Discord, WhatsApp, email) via webhooks.
• Can read and reference any file in your forked repository.
• Supports Model Context Protocol for tools.
• Can communicate with other agents in the Cocapn Fleet network.
• No telemetry, user accounts, or usage limits are imposed by the code.

How this is different

1. No central service. Every instance is independent. Prompts and memory never pass through our infrastructure.
2. Transparent and simple. The entire worker is about 1200 lines of plain JavaScript.
3. Fork-first. You own the copy. We will not push updates or breaking changes to your repository.

Honest limitation

The Cloudflare Workers free tier allows 100,000 requests per day. If you exceed this, your agent will stop responding until the next 24-hour cycle. For higher traffic, you must upgrade your Cloudflare plan.

The Fleet · Cocapn