Move API credentials to .env and set gpt-5.6-luna model

- Load OPENAI_API_KEY and OPENAI_BASE_URL from gitignored .env (repo-root, CWD-independent)
- Fix command explanation to use the configured model instead of hardcoded gpt-4o-mini
- Update default model and fallbacks to gpt-5.6-luna
- Document .env-based configuration in READMEs

.env.example

@@ -1 +1,2 @@
 OPENAI_API_KEY="your-openai-api-key-here"
+OPENAI_BASE_URL="https://api.openai.com/v1"

README.md

@@ -19,7 +19,7 @@ Arguments:
   --theme <theme>    : Set syntax highlighting theme (e.g., monokai, github-dark)
 
 Current configuration per edison.yaml:
-* Model        : gpt-4o-mini
+* Model        : gpt-5.6-luna
 * Temperature  : 0
 * Max. Tokens  : 500
 * Safety       : on
@@ -64,13 +64,27 @@ cd command-assistant
 .\install_edison.bat
 ```
 
-## OpenAI API Key Configuration
+## API Configuration
 
-Configure your OpenAI API key using one of these methods:
+Edison needs an API key (and, when using a proxy, a base URL) to talk to your model backend. The recommended way is a `.env` file in the repository root, which is gitignored and therefore never checked in:
 
-- Environment variable: `export OPENAI_API_KEY=<yourkey>` (Linux/macOS) or `$env:OPENAI_API_KEY="<yourkey>"` (Windows)
+```bash
-- Create a file at `~/.openai.apikey` with just the key
+# .env (do not commit this file)
-- Add the key to the `edison.yaml` configuration file
+OPENAI_API_KEY="your-api-key"
+OPENAI_BASE_URL="https://api.openai.com/v1"
+```
+
+Copy `.env.example` to get started:
+
+```bash
+cp .env.example .env
+```
+
+You can also provide the key using any of these methods (in order of precedence):
+
+- `OPENAI_API_KEY` / `OPENAI_BASE_URL` environment variables (or in `.env`)
+- A file at `~/.openai.apikey` containing just the key
+- The `openai_api_key` / `base_url` entries in `edison.yaml`
 
 ## Usage Examples
 

edison/README.md

@@ -80,7 +80,7 @@ The rich terminal UI provides:
 Edison can be configured via the `edison.yaml` file. Here's an example configuration:
 
 ```yaml
-model: gpt-3.5-turbo
+model: gpt-5.6-luna
 temperature: 0
 max_tokens: 500
 safety: true
@@ -88,13 +88,22 @@ safety: true
 
 ## API Key
 
-Edison requires an OpenAI API key to function. You can provide it in one of the following ways:
+Edison requires an API key to function. The recommended way is a `.env` file in the repository root (gitignored, so not checked in):
+
+```
+OPENAI_API_KEY="your-api-key"
+OPENAI_BASE_URL="https://api.openai.com/v1"
+```
+
+You can also provide it in one of the following ways:
 
 1. Environment variable: `OPENAI_API_KEY="your-api-key"`
-2. `.env` file in the same directory as the script
+2. `.env` file in the repository root
 3. `.openai.apikey` file in your home directory
 4. In the `edison.yaml` configuration file: `openai_api_key: "your-api-key"`
 
+When using a custom backend or proxy, set `OPENAI_BASE_URL` (or `base_url` in `edison.yaml`) to its endpoint.
+
 ## License
 
 MIT

edison/core/api_client.py

@@ -11,6 +11,31 @@ from edison.utils import logging_utils
 
 logger = logging.getLogger(__name__)
 
+def get_env_path():
+    """
+    Get the path to the project's `.env` file, independent of the current working directory.
+
+    Returns:
+        str: The absolute path to the `.env` file at the repository root.
+    """
+    project_root = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+    return os.path.join(project_root, ".env")
+
+def get_base_url(config):
+    """
+    Get the API base URL from the environment or config.
+
+    Prefers the `OPENAI_BASE_URL` environment variable, falling back to the
+    `base_url` value in the provided config dictionary.
+
+    Args:
+        config (dict): A dictionary containing configuration values.
+
+    Returns:
+        str or None: The API base URL, or None if not set.
+    """
+    return os.getenv("OPENAI_BASE_URL") or config.get("base_url")
+
 def get_api_key(config):
     """
     Get the OpenAI API key from various sources.
@@ -26,7 +51,7 @@ def get_api_key(config):
     Returns:
         str: The OpenAI API key
     """
-    dotenv.load_dotenv()
+    dotenv.load_dotenv(get_env_path())
 
     # Method 1: Read API key from environment variable
     # The user can set their OpenAI API key by creating a ".env" file in the same
@@ -68,6 +93,9 @@ def create_client(config):
         OpenAI: An initialized OpenAI client.
     """
     api_key = get_api_key(config)
+    base_url = get_base_url(config)
+    if base_url:
+        return OpenAI(api_key=api_key, base_url=base_url)
     return OpenAI(api_key=api_key)
 
 def call_api(client, config, query):
@@ -95,7 +123,7 @@ def call_api(client, config, query):
     try:
         # Use the modern API pattern
         response = client.chat.completions.create(
-            model=config.get("model", "gpt-4o-mini"),
+            model=config.get("model", "gpt-5.6-luna"),
             messages=[
                 {"role": "system", "content": system_prompt},
                 {"role": "user", "content": prompt}
@@ -137,7 +165,7 @@ def call_api_streaming(client, config, query, callback):
     try:
         # Use the streaming API pattern
         response = client.chat.completions.create(
-            model=config.get("model", "gpt-4o-mini"),
+            model=config.get("model", "gpt-5.6-luna"),
             messages=[
                 {"role": "system", "content": system_prompt},
                 {"role": "user", "content": prompt}

edison/edison.yaml

@@ -1,7 +1,10 @@
-model: gpt-4o-mini  # Default model for command generation
+model: gpt-5.6-luna  # Default model for command generation
 temperature: 0
 max_tokens: 500
 
+# API base URL (optional): Set via OPENAI_BASE_URL in .env, or uncomment below
+# base_url: https://api.openai.com/v1
+
 # Safety: If set to False, commands returned from the AI will be run *without* prompting the user.
 safety: True
 
@@ -27,5 +30,7 @@ ui:
   # Structured explanations: Generate more structured command explanations
   structured_explanations: True
 
-# Open AI API Key (optional): The key can aso be provided via environment variable (OPENAI_API_KEY), .env, or ~/.openai.apikey file
+# OpenAI API Key (optional): The key can be provided via .env (OPENAI_API_KEY),
-openai_api_key:
+# an environment variable, or the ~/.openai.apikey file. Prefer .env so the key
+# is not checked into version control.
+# openai_api_key: your-api-key-here

edison/ui/console.py

@@ -132,7 +132,7 @@ def handle_command_execution(client, config, command, explain=False):
             use_rich = ui_config.get("rich_formatting", True)
             
             # Generate explanation
-            explanation = get_command_explanation(client, command, structured=structured)
+            explanation = get_command_explanation(client, command, structured=structured, model=config.get("model"))
             
             if use_rich:
                 print()  # Add a blank line before the explanation

edison/ui/interactive.py

@@ -25,7 +25,7 @@ def handle_streaming_output_interactive(token):
     # Print the token without a newline and flush immediately
     print(token, end='', flush=True)
 
-def get_command_explanation(client, command, structured=False):
+def get_command_explanation(client, command, structured=False, model=None):
     """
     Get an explanation for a command.
     
@@ -33,6 +33,7 @@ def get_command_explanation(client, command, structured=False):
         client: The OpenAI client.
         command (str): The command to explain.
         structured (bool): Whether to use structured explanation format.
+        model (str): The model to use for explanation.
         
     Returns:
         str: The explanation of the command.
@@ -61,7 +62,7 @@ def get_command_explanation(client, command, structured=False):
             """
         
         response = client.chat.completions.create(
-            model="gpt-4o-mini",
+            model=model or "gpt-5.6-luna",
             messages=[
                 {"role": "system", "content": system_prompt},
                 {"role": "user", "content": command}
@@ -284,7 +285,7 @@ def interactive_mode(client, config):
                     use_rich = ui_config.get("rich_formatting", True)
                     
                     # Generate explanation
-                    explanation = get_command_explanation(client, command, structured=structured)
+                    explanation = get_command_explanation(client, command, structured=structured, model=config.get("model"))
                     
                     if use_rich:
                         print()  # Add a blank line before the explanation