When you talk to an AI assistant, it knows nothing about you. It can't see your GitHub, doesn't know what you're working on, and has no idea how you prefer to communicate. Every conversation starts from zero.
This is a problem that will only get worse. As AI becomes more integrated into our workflows — writing emails, scheduling meetings, making introductions — it needs context about who we are. Right now, that context is scattered across LinkedIn profiles, old blog posts, GitHub readmes, and whatever fragments the AI can scrape together.
There's no single, authoritative, LLM-optimized source of truth for a person.
Until now.
Meet me.txt
me.txt is an open standard for personal identity files. It's a simple markdown file that lives at the root of your personal website and tells AI systems who you are.
If you're familiar with llms.txt — the standard for making websites AI-friendly — you already understand me.txt. Same format, same simplicity, different subject: you.
# Jamie Chen
> Full-stack engineer building developer tools
## Now
- Building an open-source SDK generator
- Contributing to the OpenAPI spec working group
- Writing a CLI design guide
## Skills
- Developer experience
- API design
- CLI tooling
## Links
- [GitHub](https://github.com/jamiechen)
- [Twitter](https://twitter.com/jamiechen)
- [Email](mailto:hi@jamiechen.dev)
## Preferences
- Communication: Async-first (GitHub issues, email)
- Timezone: US Pacific (UTC-8)
- Response time: Within 24-48 hours
- Meetings: Only when async won't work
That's it. A markdown file. Human-readable, AI-native, self-authored.
The Preferences section
What makes me.txt different from a typical bio is the ## Preferences section. This is where you tell AI systems — and humans — how to interact with you:
- Do you prefer async or sync communication?
- What timezone are you in?
- How quickly do you typically respond?
- What channels do you prefer (email, Slack, GitHub issues)?
- What kind of outreach are you open to — or not?
This is especially powerful for AI-mediated introductions. When two people share their me.txt files, an AI can instantly understand not just what they do, but how they might work together effectively.
How to create yours
The fastest way to create a me.txt is with the CLI tool:
npx create-me-txt
This launches an interactive wizard that guides you through each field. When you're done, it generates a valid me.txt file ready to publish.
You can also pull data from your GitHub profile:
npx create-me-txt --github yourusername
This pre-populates your me.txt with your name, bio, top repos, and social links.
Not a developer? Just tell your AI agent:
Create a me.txt file for me following the spec at metxt.org/spec. Place it at my site root (/me.txt). Keep it concise.
Works with Cursor, Copilot, Claude, ChatGPT, or any coding agent.
Where to put it
Place your me.txt at the root of your personal website:
https://yoursite.com/me.txt
Or if you prefer:
https://yoursite.com/.well-known/me.txt
That's it. No API keys, no sign-ups, no platform lock-in. It's just a file on your domain.
What's already built
- Directory: A browseable index of me.txt files across the web. Submit yours at metxt.org/submit.
-
CLI: Generate, validate, and fetch me.txt files from the terminal with
npx create-me-txt. Pre-fill from GitHub with--github username. - API: Look up anyone's me.txt by domain at metxt.org/api/lookup.
Join the standard
me.txt is an open standard, MIT-licensed, and community-driven. The spec is simple by design — if you can write markdown, you can create a me.txt.
Ready to create yours?
npx create-me-txt
Read the full specification, browse the directory, or follow @usemetxt for updates.
The AI age needs a way for people to introduce themselves. That's me.txt.
Top comments (0)