4.1 KiB
/context/development-context.md
Development Environment Context
This document describes the preferred development workflow, tools, standards, and technology choices used by RDB Solutions Ltd.
AI coding assistants, IDE agents, and autonomous development tools should read this file before making recommendations or changes.
Primary Development Workstation
MacBook Pro (Intel)
Primary software development machine.
Typical activities:
- Application development
- Repository management
- Code reviews
- Testing
- Infrastructure administration
- LLM-assisted development
Source Control
Gitea
Primary Git platform.
Preferences:
- Gitea is the source of truth.
- Do not assume GitHub is available.
- Do not recommend GitHub-specific workflows unless explicitly requested.
- Pull Requests should be used for significant changes.
- Commit messages should be clear and descriptive.
Preferred commit style:
feat: add VM status endpoint
fix: correct Jenkins deployment script
docs: update infrastructure documentation
CI/CD
Jenkins
Primary automation platform.
Current and future responsibilities:
- Build applications
- Run tests
- Build Docker images
- Deploy services
- Run scheduled maintenance tasks
- Execute agent-generated work after review
Preferred pipeline behaviour:
- Pull code from Gitea
- Run validation
- Run tests
- Build artefacts
- Deploy to target environment
Agents should favour Jenkins automation over manual deployment procedures.
Containerisation
Docker
Preferred deployment format.
Goals:
- Consistent environments
- Easier deployment
- Simpler rollback
- Portable services
When proposing software:
Prefer:
- Docker Compose
- Single-container deployments
- Simple architecture
Avoid:
- Kubernetes unless there is a clear requirement.
Database Platforms
PostgreSQL
Preferred database platform.
Current implementation:
- Supabase-hosted PostgreSQL
Preferred characteristics:
- Relational data model
- Strong typing
- Auditable schema changes
IDEs and Editors
Current tools may include:
- VS Code
- Cursor
- Claude Code
- OpenWebUI
- Browser-based assistants
Recommendations should remain editor-agnostic where possible.
Local LLM Infrastructure
Ollama
Primary local inference platform.
Purpose:
- Local coding assistance
- Experimentation
- Agent backends
- Development support
Available Hardware
LLM Server:
- RTX 5070 Ti
- 16 GB VRAM
- 32 GB RAM
Agents should assume local inference is available.
Preferred Models
Potential models include:
- Qwen Coder
- Qwen3
- DeepSeek Coder
- Future coding-focused local models
Model selection should prioritise:
- Code quality
- Reasoning ability
- Low operational cost
- Local execution
AI-Assisted Development
RDB actively uses AI-assisted development.
Expected uses:
- Code generation
- Refactoring
- Documentation
- Testing
- Architecture review
- Infrastructure planning
Agents should produce:
- Small reviewable changes
- Clear reasoning
- Minimal surprises
Avoid:
- Massive rewrites
- Unrequested architecture changes
- Hidden behaviour
Task Management
Preferred approach:
Backlog-driven development.
Work should be decomposed into:
- Small tasks
- Independent tasks
- Reviewable tasks
Good example:
TASK-001
Create API endpoint for VM status
TASK-002
Add frontend status widget
TASK-003
Write integration tests
Avoid:
Rewrite the entire application
Documentation Standards
Projects should contain:
README.md
TASKS.md
ARCHITECTURE.md
CHANGELOG.md
/context/
Documentation should always be updated when functionality changes.
Deployment Philosophy
Preferred order:
Development → Test → Staging → Production
Production deployments should be repeatable and automated.
Avoid:
- Manual server modifications
- Untracked configuration changes
- Snowflake servers
Development Principles
Prefer:
- Simplicity
- Maintainability
- Incremental improvements
- Clear documentation
- Automation
- Reproducibility
Avoid:
- Unnecessary complexity
- Vendor lock-in
- Excessive dependencies
- Platform-specific assumptions