Gemini CLI LLM

Description

Gemini CLI LLM is a command-line interface for interacting with the Gemini API to generate content using a local large language model (LLM). This project provides a simple way to leverage the capabilities of the Gemini API directly from your terminal.

Project Summary

Core Components

• Main Application (src/main.py): CLI interface with command support for input, context, and summarization
• AI Agent (src/utils/agent.py): Implements GeminiAgent class with response generation, context awareness, and history tracking
• Configuration (src/config/settings.py): Manages API keys and base settings
• Helper Functions (src/utils/helpers.py): Handles request formatting, response parsing, and error management

Current Features

✅ Implemented:

• Basic API integration
• Context-aware prompting
• Conversation history
• Text summarization
• Token usage tracking
• Unit testing foundation

Technical Stack

• Python's requests library for API calls
• Type hints for code quality
• In-memory conversation history
• Token usage logging
• Unit tests with unittest

Usage Examples

# Basic usage
python src/main.py --input "What is machine learning?"

# With context
python src/main.py --input "What are vectors?" --context "Machine learning concepts"

# Text summarization
python src/main.py --input "Long text here" --summarize

Installation Instructions

1. Clone the repository:

   git clone https://github.com/yourusername/gemini-cli.git
   cd gemini-cli
   

2. Set up a virtual environment (optional but recommended):

   python3 -m venv venv
   source venv/bin/activate
   

3. Install the required dependencies:

   pip install -r requirements.txt
   

4. Replace the placeholder API key in src/config/settings.py with your actual Gemini API key.

Project Structure

• src/main.py: The entry point of the CLI application. It handles command-line arguments, sets up the API request to the Gemini API, and processes the response.
• src/utils/helpers.py: Contains utility functions that assist with tasks such as formatting the API request and parsing the response from the Gemini API.
• src/utils/agent.py: Implements the GeminiAgent class that handles advanced features like context-aware prompting, conversation history, and text summarization.
• src/config/settings.py: Stores configuration settings, including the API key and any other necessary parameters for connecting to the Gemini API.
• tests/: Contains test scripts to validate the functionality of the application, including unit tests for utility functions and integration tests for the main application.
• .gitignore: Specifies files and directories that should be ignored by Git, such as sensitive information like API keys and virtual environment folders.
• requirements.txt: Lists the Python dependencies required for the project, including libraries for making HTTP requests and handling JSON data.
• README.md: Comprehensive documentation for the project, explaining its purpose, setup, usage instructions, and next steps for further development.

Advanced Features

Context-Aware Prompting

The CLI now supports context-aware prompting to get more precise responses:

python src/main.py --input "What are vectors?" --context "Machine learning concepts"

Text Summarization

Summarize long texts using the summarization feature:

python src/main.py --input "Your long text here" --summarize

Conversation History

The CLI maintains a history of interactions, including:

• Prompts
• Responses
• Timestamps

Roadmap

1. Add Unit Tests

   • Write unit tests for utility functions and the main logic using pytest or unittest.
   • Status: 🚧 Under Development

2. Improve Error Handling

   • Add robust error handling for API timeouts, invalid API keys, and malformed responses.
   • Status: ❌ Not Started

3. Support Multiple Commands

   • Add commands like --history to view past token usage, --reset to clear logs, or --config to update the API key.
   • Status: ❌ Not Started

4. Add Configuration Management

   • Replace settings.py with a .env file and use python-dotenv for better security.
   • Status: ❌ Not Started

5. Add Logging

   • Use Python's logging module to log API requests, responses, and errors for debugging.
   • Status: ❌ Not Started

6. Token Usage Analytics

   • Parse the token_usage.txt file to calculate total tokens used over time and display usage statistics.
   • Status: ❌ Not Started

7. Interactive Mode

   • Add an interactive mode where users can input prompts continuously without restarting the script.
   • Status: ❌ Not Started

8. Dockerize the Project

   • Create a Dockerfile to containerize the application for easier deployment.
   • Status: ❌ Not Started

9. Add a Web Interface

   • Use Flask or FastAPI to create a web interface for the CLI.
   • Status: ❌ Not Started

10. Integrate with Other APIs

    • Extend the project to integrate with other APIs, such as summarization, translation, or sentiment analysis.
    • Status: ❌ Not Started

11. Package as a Python Library

    • Refactor the project into a Python package and add a setup.py file for distribution.
    • Status: ❌ Not Started

12. Add Documentation

    • Expand the README.md and use tools like Sphinx to generate detailed documentation.

• Status: ❌ Not Started

13. Optimize for Performance

    • Add caching for repeated prompts using functools.lru_cache or a database like SQLite.
    • Status: ❌ Not Started

14. Publish to GitHub

    • Publish the project to GitHub, add a license, and encourage contributions from the open-source community.

• Status: ❌ Not Started

New Features Status

• ✅ Context-Aware Prompting
• ✅ Conversation History
• ✅ Text Summarization
• ✅ Enhanced Error Handling
• 🚧 Retry Mechanism (max_retries=3)

AI Memory & Persona Roadmap

1. Persistent Conversation Storage 🔄

• Implement PostgreSQL database to store conversations

  -- Create extensions for vector search capabilities
  CREATE EXTENSION IF NOT EXISTS vector;
  
  -- Create enum for persona types
  CREATE TYPE persona_type AS ENUM ('gym_coach', 'teacher', 'programmer', 'default');
  
  -- Create conversations table
  CREATE TABLE conversations (
      id SERIAL PRIMARY KEY,
      timestamp TIMESTAMPTZ DEFAULT CURRENT_TIMESTAMP,
      user_input TEXT NOT NULL,
      ai_response TEXT NOT NULL,
      active_persona persona_type DEFAULT 'default',
      context_tags TEXT[],
      embedding vector(1536),  -- For semantic search
      metadata JSONB
  );
  
  -- Create index for efficient context search
  CREATE INDEX idx_conversations_timestamp ON conversations(timestamp);
  CREATE INDEX idx_conversations_persona ON conversations(active_persona);
  CREATE INDEX idx_conversations_embedding ON conversations USING ivfflat (embedding vector_cosine_ops);

• Required PostgreSQL setup:

  sudo apt install postgresql postgresql-contrib
  sudo systemctl start postgresql
  sudo -u postgres createdb gemini_cli

• Dependencies to add in requirements.txt:

  psycopg2-binary>=2.9.9
  sqlalchemy>=2.0.0
  alembic>=1.12.0
  

• Status: ❌ Not Started

2. Persona Management System 🎭

• Create persona configuration system
• Add persona commands:

  python src/main.py --set-persona "gym_coach"
  python src/main.py --list-personas

• Status: ❌ Not Started

3. Context Window Management 🪟

• Implement sliding context window
• Add context pruning strategies
• Status: ❌ Not Started

4. Memory Types Implementation 🧠

• Short-term memory (current session)
• Long-term memory (persistent storage)
• Working memory (active context)
• Status: ❌ Not Started

5. Memory Search & Retrieval 🔍

• Implement vector embeddings
• Add memory commands:

  python src/main.py --search "previous workouts"
  python src/main.py --context-depth deep

• Status: ❌ Not Started

6. Conversation Threading 🧵

• Add thread management:

  python src/main.py --new-thread "workout_plan"
  python src/main.py --switch-thread "diet_advice"

• Status: ❌ Not Started

7. Memory Consolidation 📝

• Periodic memory summarization
• Knowledge base updates
• Status: ❌ Not Started

8. User Preference Learning 👤

• Track user interactions
• Adapt persona behavior
• Status: ❌ Not Started

9. API Integration 🔌

• Enhance GeminiAgent class
• Add memory management methods
• Status: ❌ Not Started

10. Performance Optimization ⚡

• Implement caching
• Optimize database queries
• Status: ❌ Not Started

Let's get to work. There's a lot to be done.

---