docs/compose: update docker-compose and .env for image features

Signed-off-by: Hrushikesh Patil <[email protected]>

Author: nutanix-Hrushikesh

.env.ollama

@@ -2,3 +2,4 @@ CHAT_MODEL=qwen2.5:0.5b
 THINKING_MODEL=qwen3:1.7b
 COMPLETION_MODEL=qwen2.5:0.5b
 EMBEDDINGS_MODEL=all-minilm:33m
+IMAGE_GENERATION_MODEL=dall-e-2

cmd/aigw/README.md

@@ -56,9 +56,11 @@ Here are values we use for Ollama:
      ```
 
    - MCP (Model Context Protocol) tool call:
+
      ```bash
      docker compose run --rm mcp
      ```
+
      This calls the kiwi MCP server through aigw's MCP Gateway at `/mcp`.
 
    - Image generation:
@@ -72,14 +74,15 @@ Here are values we use for Ollama:
          -X POST http://localhost:1975/v1/images/generations \
          -H "Authorization: Bearer unused" \
          -H "Content-Type: application/json" \
-         -d '{"model":"gpt-image-1","prompt":"A watercolor painting of a red fox in a birch forest","size":"512x512"}' \
+         -d '{"model":"dall-e-2","prompt":"A watercolor painting of a red fox in a birch forest","size":"512x512"}' \
          | jq -r '.data[0].b64_json' | base64 -d > image.png
        ```
 
 4. **Create embeddings**:
 
    The `create-embeddings` service uses `curl` to send an embeddings request
    to the AI Gateway CLI (aigw) which routes it to Ollama.
+
    ```bash
    docker compose run --rm create-embeddings
    ```

cmd/aigw/docker-compose.yaml

@@ -22,7 +22,7 @@ services:
     image: alpine/ollama
     container_name: ollama-pull
     environment:
-      OLLAMA_HOST: localhost:11434 # intentionally not 127.0.0.1
+     OLLAMA_HOST: localhost:11434 # intentionally not 127.0.0.1
     env_file:
       - ../../.env.ollama
     entrypoint: sh
@@ -152,6 +152,6 @@ services:
           -X POST http://aigw:1975/v1/images/generations \
           -H "Authorization: Bearer unused" \
           -H "Content-Type: application/json" \
-          -d "{\"model\":\"$$CHAT_MODEL\",\"prompt\":\"A watercolor painting of a red fox in a birch forest\",\"size\":\"256x256\"}"
+          -d "{\"model\":\"$$IMAGE_GENERATION_MODEL\",\"prompt\":\"A watercolor painting of a red fox in a birch forest\",\"size\":\"256x256\"}"
     extra_hosts:  # localhost:host-gateway trick doesn't work with aigw
       - "host.docker.internal:host-gateway"