MD = Managing Director, someone who’s overseeing a project and coordinating multiple parties.

Similarly, you are overseeing multiple projects and systems, email, calendar, etc.

And how you operate those systems follows rules.

Similarly, when you are instructing and framing instructions to ChatGPT on how you want it to complete a task and use tasks you give it specific rules.

What does that look like from a coding use case where you might be working with various tools and providing context on how you want a system to execute?

Enter the previous state; individual context prompts for each tool, e. g., instructions for cursor, Claude, Gemini.

Now the new state: add project overview, build and test commands, code style guidelines, and any other relevant instructions. Then include commit messages, pull request guidelines, and anything else a new teammate would need to know and add that into an Agents.md file.

Now you have a managing director level context for your various agents to pull from.

And the ship stays on track.