Give it standing instructions

You shouldn't have to repeat who you are and how you like to work at the top of every chat. Every tool has a place for instructions that stay in force. They just call it different things.

The habit

Write your standing instructions once: who you are, what you're working on, the tone you want, and what to never do. Keep them short. A long list gets skimmed by the model the same way it gets skimmed by a person.

Side by side

• Claude: a Project has its own instructions that apply to every chat in it. In Claude Code, the same job is done by a CLAUDE.md file in your project; /init writes a first one.
• ChatGPT: a project has project instructions that "apply across its chats".
• Grok: we didn't find standing instructions in xAI's Grok docs, so paste your short version at the top of the chat.
• Cursor: project rules live in .cursor/rules as .mdc files, and Cursor also reads an AGENTS.md file in your project. You can add your own User Rules too.

One file, many tools

AGENTS.md is worth knowing: Cursor reads it, OpenAI's Codex reads it, and Claude Code reads it when there's no CLAUDE.md. If you work in more than one coding tool, put your project's rules there. If you also keep a CLAUDE.md, add the line @AGENTS.md to it so Claude Code reads both.

Checked against each vendor's docs on 2026-10-14: Claude Projects, Claude Code memory, ChatGPT projects, Grok FAQ, Cursor rules, Codex AGENTS.md.