Use plan.md to Coordinate Work Across Parallel Claude Sessions
Running multiple Claude sessions in parallel is powerful, but without coordination they can duplicate work, conflict on files, or undo each other's changes. A shared plan.md in the project root solves this.
Each session reads the plan before acting, and updates it before handing off:
# Feature: Checkout Flow
## In Progress
- Session A: payment form validation (src/components/PaymentForm.vue)
- Session B: writing tests for OrderService (tests/Feature/OrderTest.php)
## Completed
- [x] Cart summary component
- [x] Address form with validation
## Blocked
- Email receipt — waiting on SMTP credentials
## Decisions
- Using Stripe Elements for card input (no raw card data on our server)
- Skipping PayPal for v1
Before starting work, each Claude session reads the plan:
Read plan.md. I'm Session B. What's left for me to work on that doesn't
conflict with what's in progress?
After completing a step, the session updates the plan:
Mark "writing tests for OrderService" as complete in plan.md and
note any constraints the next session should know about.
The plan.md acts as the shared memory that all sessions — and all future context windows — can rely on. Unlike chat history, it persists on disk and is readable by any session at any time.
With a shared plan.md, parallel sessions become a coordinated team rather than independent workers who don't know what the others are doing.
via @ShaneCollinsAI
Log in to leave a comment.
Set up Claude Code as an automated reviewer in your CI pipeline — on every pull request, it reads the diff, checks for bugs, security issues, missing tests, and convention violations, then posts its findings as a PR comment. Your human reviewers get a head start because the obvious issues are already flagged before they look.
Before deploying, tell Claude to read your project — migrations, environment variables, queue workers, scheduled tasks, caching, third-party integrations — and generate a deployment checklist that's specific to your app. Not a generic "did you run migrations?" list, but one that knows YOUR infrastructure and catches the things YOUR deploy can break.
Instead of writing a README from memory or copying a template, tell Claude to read your project and generate one that's actually accurate — real setup instructions from your config, real architecture from your directory structure, real API examples from your routes, and real prerequisites from your dependency files.