Working with AI Tools
Omaship is designed to work seamlessly with AI coding assistants like Claude Code and Cursor.
Why Rails Works Well with AI
- Convention over configuration โ Predictable file locations
- Two decades of patterns โ Extensive training data
- Consistent structure โ Every Rails app follows the same layout
AGENTS.md
Every Omaship project includes an AGENTS.md file at the root (CLAUDE.md is symlinked to it for tool compatibility). This provides project-specific context to AI tools:
- Project architecture overview
- Key conventions and patterns
- Important files and their purposes
- Common tasks and how to accomplish them
Update this file as your project grows. The more context you provide, the better AI tools perform.
You can also install agent skills to give your AI tools procedural knowledge about specific frameworks and tools used in your project.
Tips for Effective AI Coding
Be Specific About Location
Less effective: "Add a contact form"
More effective: "Add a contact form to app/views/pages/contact.html.erb that submits to ContactsController#create"
Reference Existing Patterns
"Add a new model following the same pattern as app/models/user.rb"
Use Rails Commands
AI tools understand Rails generators:
"Run 'rails generate model Product name:string price:decimal' and create a ProductsController with index and show actions"
ast-grep Rules
Omaship includes ast-grep rules that enforce code quality patterns. AI tools respect these rules when generating code. See sgconfig.yml for the full list.
Getting Help
If an AI tool gets stuck, provide more context:
- Share relevant file contents
- Describe the expected behavior
- Point to similar working code
Tip: The more you treat AI tools as junior developers who need context, the better results you'll get.