Start with about-me.md

The easiest AI context system is usually not a system yet.

It is a small folder of plain text: who you are, what you are working on, what is still undecided, what already changed, and what you want an assistant to do with the material.

root fits that middle step. You can keep the files as normal Markdown, drag or import them into root, tag the related notes, save repeat instructions as Snippets, and use Context Handoff when a selected packet is ready for ChatGPT, Claude, Gemini, or another AI tool.

Start with one identity file

Make about-me.md first. Not a memoir. Not a permanent profile. Just a small context file that helps repeated AI conversations start with the basics already clear.

It might hold:

• what you do
• how you like help structured
• recurring tone preferences
• constraints you repeat often
• things an assistant should not assume

Keep it plain. A useful about-me.md can be a short list that takes five minutes to write and one minute to revise later.

Add a few working files

After that, add only the files you already have a use for. A simple folder might include:

• current-focus.md for what matters this week
• open-questions.md for decisions that still need thought
• decisions.md for what already changed and why
• todo.md for plain checklist lines
• notes.md for rough fragments that do not need a title yet
• links.md for references with a sentence of context
• draft.md for the current thing you are trying to write
• meeting-notes.md for raw notes before they become follow-ups
• project-brief.md for the short version of the work

In the current Chrome beta, root supports importing or dragging one or more .md or .txt files into the workspace. That makes this setup low-ceremony: ordinary files can stay ordinary until you want to work with them together.

Keep existing tool files when they help

Some people already have AI or development-adjacent Markdown files near their work. root does not require any of these special names, but it can work with them because they are still just portable Markdown files.

Recognizable examples include:

• README.md for basic background
• CHANGELOG.md for what changed over time
• CONTRIBUTING.md for rules and expectations
• AGENTS.md for agent-facing instructions
• CLAUDE.md for Claude-specific working notes
• GEMINI.md for Gemini-specific working notes

Treat these as examples, not requirements. If you already have them, they can become useful context. If you do not, start with about-me.md and the few working files that match your real work.

Use tags instead of a heavy visual dashboard

The point is not to build a visual command center for your life. root stays light on purpose: tags group related context when it is useful, without turning your working files into another system to maintain.

You might tag about-me.md, current-focus.md, and project-brief.md with #launch. Or tag links.md, meeting-notes.md, and open-questions.md with #support. The tag is enough to gather the slice later.

A simple tag is enough to bring related context back together when you need it.

Save repeat instructions as Snippets

The context files are the material. Snippets are the repeat instructions you reach for again and again. Save the reusable move once, then add it only when the packet is ready to leave root.

Good starter Snippets include:

• summarize this context
• extract open tasks
• turn this into a support reply
• write a product update
• make this clearer without making it formal

This keeps reusable instructions close without turning the main workspace into a prompt archive.

Use Context Handoff for the actual AI chat

The handoff is where the setup earns its keep. Instead of pasting a whole folder into an assistant, choose the current item, selected notes, a tagged slice, pinned context, or Open Tasks. Add a saved Snippet if the packet needs framing, then inspect what is about to leave.

That packet can go into ChatGPT, Claude, Gemini, browser AI, email, a doc, or another tool. The important part is that the packet is deliberate. The assistant gets the useful working context, not every nearby file just because it happened to be in the folder.

Private by default, honest at the boundary

root's local workspace data is private by default. The notes stay in your local workspace unless you choose to copy, export, or share them.

Once you paste or share a Context Handoff packet into an AI chat, that tool may see the context you gave it. That is why the boundary matters: root helps you decide what belongs in the packet before it leaves.

A 10-minute starting folder

If you want the smallest useful version, make this folder:

• about-me.md
• current-focus.md
• open-questions.md
• todo.md
• links.md

Drag those files into root. Tag the ones that belong together. Save one or two Snippets for repeat instructions. Then use Context Handoff the next time you want AI help on the current slice.

That is enough to start. The structure can grow only when the work asks for it.