Update README and install script: add installation instructions and create dependency installation script

README.md

@@ -18,6 +18,12 @@ cd _llm
 pip install -e .
 ```
 
+Alternatively, after cloning, you can install all dependencies (including those from GitHub) using the provided script:
+
+```bash
+bash install_deps.sh
+```
+
 ## Dependencies
 
 This package requires:

_llm/llm.py

@@ -3,14 +3,12 @@ import base64
 import re
 import traceback
 from typing import Literal, Optional
-import requests
 import tiktoken
 from ollama import (
     Client,
     AsyncClient,
     ResponseError,
     ChatResponse,
-    Tool,
     Options,
 )
 
@@ -371,7 +369,36 @@ class LLM:
         force_local: bool = False,
     ):
         """
-        Generate a response based on the provided query and context.
+        Attempts to generate a response using a remote API first, then falls back to
+        local Ollama if the remote call fails or if force_local is True.
+
+        Args:
+            query (str, optional): The main query or prompt for generation.
+            user_input (str, optional): Alternative user input if query is not provided.
+            context (str, optional): Additional context to include in the generation.
+            stream (bool, optional): Whether to stream the response. Defaults to False.
+            tools (list, optional): List of tools to make available for the model.
+            images (list, optional): List of images to include in the request.
+            model (Literal["small", "standard", "vision", "reasoning", "tools"], optional):
+                The model type to use. Defaults to "standard".
+            temperature (float, optional): Temperature parameter for generation randomness.
+                Uses instance default if not provided.
+            messages (list[dict], optional): Pre-formatted message history.
+            format (optional): Response format specification.
+            think (bool, optional): Whether to enable thinking mode. Defaults to False.
+            force_local (bool, optional): Force use of local Ollama instead of remote API.
+                Defaults to False.
+
+        Returns:
+            The generated response. Type varies based on stream parameter and success:
+            - For streaming: Returns stream reader object
+            - For non-streaming remote success: Returns response message object
+            - For local fallback: Returns local Ollama response
+            - For complete failure: Returns error message string
+
+        Raises:
+            Prints stack trace for exceptions but doesn't propagate them, instead
+            returning error messages or attempting fallback to local processing.
         """
         model = self._prepare_messages_and_model(
             query, user_input, context, messages, images, model

install_deps.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+# Install required dependencies for llm_client
+
+pip install --upgrade pip
+pip install git+https://github.com/lasseedfast/env_manager.git
+pip install git+https://github.com/lasseedfast/colorprinter.git
+
+# Optionally install other requirements
+pip install requests tiktoken ollama
+
+echo "All dependencies installed."