Problem

This epic is implementation of Remote engine

Success Criteria

• Cortex supports remote engine for using remote provider (openai, anthropic, ...)
• Cortex supports engine managements for remote engine
• Cortex supports model managements for remote model (gpt-4o, claude, ...)
• Cortex supports API for Chat Completions, Get models and route request to correct remote provider

Tasklist

• Implement jinja2 parser and chat completion requests route
• Implement Get models request route
• Implement Request Transform
• Implement Response Transform
• Implement Engine management
  • Engine table in cortex.db
  • Upsert Remote engine (name & variant would be unique as composite key)
  • Get Remote engine by Id
  • Get Remote engine by name and variant (optional)
  • Delete Remote engine
  • Update API /v1/engines to support remote engine
• Implement model management (cortex.cpp)
  • Add remote model.yml
  • Update Remote model.yml
  • List, Get remote models
  • load, unload model
  • chat completion with non stream request
  • chat completion with stream request
  • Integrate with new engine implementation
• Try using transform template with Claude model
• Support CLI for remote engines
• Docs