# Copilot Prompt for Rust Image Manipulation API

You are working on a Rust API project built with:
- Axum 0.8.8
- SeaORM 1.1.20 with PostgreSQL
- Tokio full runtime
- tower-http trace

The project already has an account system. Do not redesign it unless asked.

## Main Goal
Build an image manipulation API that:
- supports `avif`, `png`, `jpg`, `jpeg`, `webp`
- accepts `width`, `height`, `format`, `quality`, and `crop`
- supports crop modes like `cover` and other common crop strategies
- uses time-limited access tokens for protected image operations
- uses Redis for token validation and temporary access control

## Important Constraints
- Use idiomatic Rust
- Follow Axum conventions
- Keep business logic in services
- Keep DB and Redis logic in repositories
- Use SeaORM for database access
- Validate all inputs
- Return structured JSON errors
- Do not panic in handlers
- Do not expose internal file paths or secrets

## Implementation Guidance
When generating or modifying code:
1. Split code by module:
   - controller/handler
   - service
   - repository
   - dto
   - model
   - processor

2. For image processing:
   - decode input image
   - validate requested dimensions and format
   - apply crop/resize strategy
   - convert to requested output format
   - encode with requested quality
   - return the transformed image

3. For token access:
   - read token from request
   - verify token in Redis
   - reject expired/invalid tokens
   - proceed only if valid

4. For API design:
   - keep routes clean and versioned if needed
   - use consistent request/response shapes
   - add tracing/logging where useful

## Default Behaviors
- default format: `webp`
- default crop: `cover`
- default quality: `80`
- if only width or only height is provided, preserve aspect ratio when possible

## Output Expectations
When asked to generate code:
- provide complete Rust code snippets
- include file names
- keep code ready to integrate
- minimize unnecessary abstractions
- match the existing codebase style

## If there is ambiguity
Prefer:
- safe validation over permissive behavior
- explicit errors over silent fallback
- modular design over monolithic handlers
- Redis-backed token checks over in-memory shortcuts