Brainfile / CLAUDE.md Configurations
15 Production-Ready Configurations

Stop Writing CLAUDE.md
Files From Scratch

Every Claude Code project needs a CLAUDE.md. Writing one that actually works takes hours of iteration. We've done it 15 times — for every role.

Free by email: a practical Claude Code tip every Tuesday and Thursday, plus a Sunday recap.

Start your 30-day free trial — $99/mo 3 Free Configurations
What is CLAUDE.md?

The file that makes Claude Code actually know your business

When you run claude in your project directory, Claude Code automatically reads CLAUDE.md in the root. This file is your persistent context — it tells Claude your project structure, your coding style, your domain expertise, and what to avoid. Without it, every session starts from zero.

CLAUDE.md markdown
# Your project instructions — loaded by Claude Code every session # Project Overview B2B SaaS — accounts receivable automation for mid-market companies. Stack: Next.js 14, TypeScript, PostgreSQL, Prisma, Stripe, Resend. Production deploys via Vercel. Staging at staging.acme.com. # Code Style - TypeScript strict mode. No `any`. Prefer type assertions over casting. - React components: functional, hooks only. Class components = refactor target. - API routes: Zod validation at boundary. Never trust raw req.body. - Error handling: structured errors with codes, never bare strings. # Architecture - Feature folders: src/features/{feature}/components, hooks, api, types - Shared components: src/components/ui/ (shadcn only, no custom) - DB models: src/db/schema.ts (single file, Drizzle schema) # Critical Laws - NEVER commit .env files. NEVER expose STRIPE_SECRET_KEY in client code. - Migration files are immutable once deployed to staging. - All Stripe webhook handlers must be idempotent.
3–5 hrs
DIY from scratch
→
15 min
With a configuration
A good CLAUDE.md is 300–800 lines. It covers your stack, architecture patterns, coding style, testing approach, security rules, domain knowledge, and what Claude should never do. Writing this from scratch — and iterating until it actually works — takes most developers 3–5 hours and multiple sessions. Start with an expert-crafted configuration and customize.
What's in a good CLAUDE.md?

The 8 sections that separate useful from useless

Most CLAUDE.md files fail because they're too vague. "Please write clean code" does nothing. Here's what actually works:

What works

  • Specific architecture rules ("feature folders, not flat components")
  • Explicit DO NOT instructions ("never mock the database in tests")
  • Domain context ("this is B2B SaaS, not consumer — enterprise security matters")
  • Tech stack specifics with version constraints
  • Your testing philosophy with concrete examples
  • Critical security and business rules that can't break
  • Your preferred patterns (hooks vs class, SQL vs ORM)
  • How to handle errors, logging, and observability

What doesn't work

  • "Write clean, well-documented code"
  • "Be careful about security"
  • "Follow best practices"
  • No architecture description
  • Generic style guidelines with no examples
  • No domain context (Claude doesn't know your business)
  • Missing DO NOT rules (Claude will repeat bad patterns)
  • No mention of your actual tech stack
Configuration Library

15 profession-specific configurations — ready to customize

Each configuration is 300–600 lines. Structured sections for every aspect of your work. Customize the specifics, keep the structure.

💻
Developer
Senior SWE. Architecture, stack, testing, debugging, code style, security rules, CI/CD.
Free Sample
🚀
Founder
Startup CEO building with Claude. GTM strategy, fundraising, product decisions, investor comms.
Free Sample
📈
Trader
Active trader. Strategy rules, risk management, market regimes, backtesting standards.
Free Sample
📣
Marketer
Growth lead. Brand voice, channel strategy, analytics setup, content standards, reporting.
Pro
🏢
Executive
C-suite. Strategic frameworks, board prep, financial oversight, decision-making protocols.
Pro
⚙️
DevOps / SRE
Infrastructure engineer. IaC standards, incident response, cost management, security hardening.
Pro
📋
Product Manager
Senior PM. PRD writing, prioritization, customer research, stakeholder management, roadmaps.
Pro
🧬
Data Scientist
ML engineer. Experiment design, model pipelines, statistical rigor, production deployment.
Pro
⚖️
Lawyer
Corporate attorney. Contract drafting, due diligence, research standards, client management.
Pro
✍️
Content Creator
Newsletter + YouTube. Editorial standards, brand voice, growth strategy, content repurposing.
Pro
💼
Consultant
Management consultant. Hypothesis structuring, deck building, analysis standards, client engagement.
Pro
🎨
Designer
Product designer. Design systems, accessibility standards, UX research, engineering handoffs.
Pro
🏠
Realtor
Real estate agent. Listings, buyer management, negotiation, lead generation, compliance.
Pro
🔬
Researcher
Academic PI. Paper writing, grant proposals, lab management, mentoring, statistical rigor.
Pro
🏢
Agency Owner
Digital agency. Client management, proposals, pricing strategy, team scaling, automation.
Pro
Common Questions

FAQ

Is CLAUDE.md only for Claude Code (CLI)?
Primarily, yes. Claude Code reads CLAUDE.md automatically when you run claude in a project directory. But the configurations work equally well as custom instructions in claude.ai Projects, or pasted at the start of any AI conversation. They're structured markdown — portable across Claude, ChatGPT, Gemini, and Cursor.
How is this different from generic prompt configurations?
Generic prompts are one-off instructions for a single task. A CLAUDE.md is persistent context that loads every session, covering your entire role — not just "write me an email." It's your AI's permanent background knowledge about you, your work, your standards, and what it should never do. Depth and specificity are what make these valuable.
Why would I pay $99/mo when I can write my own?
You can. It takes 3–5 hours the first time, and several iterations before it actually works well. Beyond that, Claude updates frequently — what works today may need tuning as the model changes. The subscription includes a weekly freshness feed that tells you what changed in Claude Code, so your configuration stays current.
Do I need Claude Code (the CLI) to use these?
Brainfile is built for Claude Code: the files arrive through a terminal pull, and the Autopilot protocol and /start interview run in Claude Code. The CLAUDE.md text is plain markdown, so you can paste parts of it into Claude.ai Projects or another tool by hand; Brainfile ships no ChatGPT or .cursorrules format.
Can I customize the configurations?
That's the whole point. Each configuration ships with clear [PLACEHOLDER] sections — fill in your actual name, company, tech stack, current projects, and preferences. The structure is the hard part; we've done that. The customization takes 15–30 minutes and makes Claude dramatically more useful to you specifically.
What does the monthly subscription actually include?
Pro: all 15 role packs, plus the Autopilot protocol, 10 pattern files and the weekly freshness feed (we use these ourselves daily). Basic ($19/mo) has everything except the role packs. As long as you subscribe, the feed tells you when Claude Code changes in ways that affect what works.
Get Started Today

15 role packs on Pro.
Weekly freshness feed. Cancel anytime.

Start the 30-day trial free, no card. Upgrade when you see how much faster Claude Code works with a real CLAUDE.md.

$99 /month
Start your 30-day free trial Try 3 Free First

30-day money-back guarantee · No setup required · Instant access

Not ready to buy? Stay in the loop.

New CLAUDE.md configurations, context engineering tips, and best practices. Free: a Claude Code tip every Tuesday and Thursday, plus a Sunday recap.

No spam. Unsubscribe anytime.