Add a New Translator

Add a new LLM provider translator to Envoy AI Gateway. This is the most common contribution type. A translator converts requests from one API schema (e.g., OpenAI) to another (e.g., AWS Bedrock) and translates responses back.

Before starting: Read aigw-contrib-go-style — performance patterns and JSON handling are the most common reasons translator PRs get rejected.

Phase 1: Understand the Interfaces

Study an existing translator to understand the pattern. Start with a simple one like openai_openai.go (passthrough), then look at a complex one like openai_awsbedrock.go.

Key interfaces your translator must implement:

1. Request translation: (sourceBody []byte) → (targetBody []byte, error)
2. Response translation: (targetBody []byte) → (sourceBody []byte, error)
3. Streaming chunk translation: (targetChunk []byte) → (sourceChunk []byte, error)

The translator is generic over [ReqT, RespT, RespChunkT] type parameters.

Phase 2: Create the Translator File