Using Claude to Speed Up Goalixa Development

Published: 2026-03-12 | Section: Software Engineering | Author: Amirreza Rezaie

Goalixa is close to production, but I need faster backend/frontend improvements. I can’t always wait for another developer, so I want to use Claude effectively for reviews and small features.

Problems today

• Every Claude session needs long context because there is no good CLAUDE.md per service.
• Auto-generated CLAUDE.md files were low quality and miss real project details.
• When I ship AI-written code, I know it may need refactor, but I need quick fixes and small features safely.

Plan to make Claude productive

• Create a concise CLAUDE.md in each microservice with:
  • What the service does and main endpoints/features
  • Stack, build, test, lint/format commands
  • Run/dev steps and sample .env
  • Deploy info: image name, tag convention, kustomize/helm path
  • Contracts: DB/queues/external APIs links (OpenAPI/Proto)
  • High-risk areas and known tech debt
  • 2–3 sample prompts for common tasks
• Add a repo-level CLAUDE.md mapping all services and shared tooling.
• Standard session intro prompt: “Use CLAUDE.md, follow coding standards, don’t touch secrets, run tests with <cmd>.”
• Keep AI changes small: one feature/fix per request, demand a test plan and self-review checklist.

How Claude fits Goalixa workflow

• Use it for:
  • Small API/FE changes with clear acceptance criteria
  • Refactoring low-risk modules
  • Writing tests and docs
• Avoid:
  • Secret handling, migrations without review, and broad rewrites.

Next steps

• Write/refresh CLAUDE.md for each service before the next dev sprint.
• Add repo-level CLAUDE.md so prompts are shorter.
• Enforce “small change + tests” when asking Claude for code.