How do I share context between Claude Code, Codex and Cursor?
By Soumen Sarkar · October 2026
Short answer: Keep the team's context in one place that every tool can reach over MCP, and let each tool's own file (CLAUDE.md, AGENTS.md, Cursor rules) hold only the stable instructions. Decisions, assumptions and work in progress change too fast to live in files that each tool reads separately.
Why isn't one instruction file enough?
Each tool reads its own file. Claude Code reads CLAUDE.md, Codex and many others read AGENTS.md, and Cursor has its rules. Teams using more than one tool end up keeping two or three copies of the same instructions, and the copies drift.
That's fine for things that rarely change: how to run the tests, which folders not to touch, the house style. It breaks down for things that change every hour:
- what a teammate's agent is changing in the billing code right now;
- the decision made this morning about how refunds work;
- the assumption an agent made last night that someone corrected at 8am.
No one is going to paste those into three files, and no agent can read a decision that only lives in a chat thread.
What should go where?
- In the instruction files: commands, conventions, boundaries and a line telling the agent where the team's live context is and when to read it. Keep CLAUDE.md and AGENTS.md short, and have one import the other so there's a single source.
- In one shared source every tool can reach: decisions with their reasons, assumptions, work in progress and handoffs. MCP, the Model Context Protocol, is the common way to do this: Claude, Claude Code, Codex, Cursor and most other agents can all connect to the same MCP server.
For more on the split, see what goes in CLAUDE.md, AGENTS.md and team memory.
How do agents use the shared source?
The best set-up doesn't rely on the agent remembering to look. The first time an agent touches an area, it's told what's changing there, what was decided and what was assumed: an area brief. When it makes a call, the call is recorded where every other tool will find it.
For example, a developer using Codex decides that invoices are generated at the end of the month. An hour later, a teammate's Claude Code session opens Invoices · lib/invoices and is told that decision before it changes anything. Neither of them had to copy anything anywhere.
Campfyrr is one shared source like this: one MCP connection for every tool your team uses, plus hooks for Claude Code that brief each session automatically. Start with connecting Claude, Claude Code and Codex.
Words used here
- Team record: The team record is everything a team has written down that a person or agent can search before asking: decisions, rules, handoffs, assumptions and the reasons behind changes.
- Area brief: An area brief is what a person or agent is told the first time they touch a part of the code: what's changing there now, what was decided, and what was tried.
- Decision: A decision is a call made in the work, recorded with its reason at the moment it is made, so it travels to everyone who touches the same ground.
The principle behind this: Answer it yourself: Look first. If the answer isn't there, assume it or decide it.
Move as one at agent speed. Nobody waits. Add Campfyrr to Claude: it briefs every agent before it touches the code, keeps assumptions and handoffs where the team can see them, and never takes your code (what it keeps).