Hermes: Your AI-Powered Command-Line Assistant

Hermes is a powerful command-line tool that allows you to interact with various AI models, process files, and enhance your productivity. Named after the messenger god in Greek mythology, Hermes facilitates seamless communication between you and AI models.

What is Hermes?

Hermes is a versatile application that enables you to:

• Chat with different AI models directly from your command line
• Process and analyze files using AI
• Incorporate AI-generated responses into your workflow

Whether you're a developer, researcher, or anyone looking to leverage AI in their daily tasks, Hermes provides an easy-to-use interface to harness the power of various AI models.

Installation

Follow these steps to install Hermes:

1. Open your terminal or command prompt.

2. Clone the Hermes repository:

   git clone https://github.com/KoStard/HermesCLI
   

3. Navigate to the Hermes directory:

   cd HermesCLI
   

4. Install Hermes in editable mode:

   pip install -e .
   

Configuration

After installation, you need to set up your configuration file. This file will store your API keys and default settings.

1. Create a configuration directory:

   • On macOS and Linux:

     mkdir -p ~/.config/hermes
     

   • On Windows:

     mkdir %USERPROFILE%\.config\hermes
     

2. Create a configuration file named config.ini in the directory you just created:

   • On macOS and Linux:

     touch ~/.config/hermes/config.ini
     

   • On Windows:

     type nul > %USERPROFILE%\.config\hermes\config.ini
     

3. Open the config.ini file in a text editor and add your configuration:

   [BASE]
   model = claude-sonnet-3.5
   
   [ANTHROPIC]
   api_key = your_anthropic_api_key
   
   [GEMINI]
   api_key = your_gemini_api_key
   
   [OPENAI]
   api_key = your_openai_api_key
   
   [DEEPSEEK]
   api_key = your_deepseek_api_key
   model = deepseek-coder

   Replace your_*_api_key with your actual API keys.

What are API Keys?

API keys are unique identifiers that authenticate your requests to AI services. They are like passwords that allow you to access and use these services. Each AI provider (like Anthropic, Google, OpenAI) has its own method for generating and managing API keys.

Basic Usage

Here's how to use Hermes:

1. Open your terminal or command prompt.

2. To start a conversation with the default AI model:

   hermes
   

3. To use a specific model and include files for context:

   hermes --model claude-sonnet-3.5 document1.txt document2.txt
   

4. After running the command, you'll be prompted to enter your message or question.

Command-line Options

Hermes supports various command-line options:

• --model: Specify the AI model to use (e.g., claude-sonnet-3.5, gemini, openai)
• --prompt: Provide an initial prompt directly
• --prompt-file: Use a file containing the initial prompt
• --append or -a: Append the AI's response to a specified file
• --update or -u: Update a specified file with the AI's response
• --pretty: Display the output with markdown formatting
• --image: Include an image file in the context (can be used multiple times)
• --text: Add extra text to the context (can be used multiple times)
• --url: Include content from a URL in the context (can be used multiple times)
• --clipboard: Include the content of the clipboard in the context

Examples

1. Summarize a document:

   hermes --model claude-sonnet-3.5 README.md --prompt "Summarize this document in one paragraph"
   

2. Ask a question and save the response:

   hermes --prompt "What is the capital of France?" --append answer.txt
   

3. Use an image in your query:

   hermes --model gemini --image photo.jpg --prompt "Describe this image"
   

4. Pipe input and output:

   echo "Tell me a fun fact" | hermes > fun_fact.txt
   

5. Use clipboard content in your query:

   hermes --clipboard --prompt "Analyze the code in my clipboard"
   

Supported Models

Hermes supports a wide variety of AI models:

• Claude (Anthropic): claude-sonnet-3.5
• Gemini (Google)
• GPT-4 and other models (OpenAI)
• Bedrock models (AWS):
  • bedrock/sonnet-3
  • bedrock/sonnet-3.5
  • bedrock/opus-3
  • bedrock/mistral
• Ollama (for local models)
• Groq
• SambaNova
• DeepSeek
• OpenRouter:
  • openrouter (default model)
  • openrouter/perplexity
  • openrouter/o1-mini
• Vertex (Google):
  • vertex/llama3-405b-instruct-maas

Each model may have specific setup requirements. Refer to the documentation of each AI provider for detailed instructions on how to use their models.

Vertex Configuration

To use the Vertex model, you need to set the project_id in the [VERTEX] section of your config.ini file. You also need to have the gcloud CLI installed and authenticated to access the Vertex AI service.

gcloud auth application-default login

This command is used to authenticate with your Google Cloud account and set up the default credentials.

Configuration File (config.ini)

The config.ini file is used to store your API keys and default settings. Here's a more detailed explanation of the possible sections and attributes:

[BASE]
model = claude-sonnet-3.5  # Default model to use

[ANTHROPIC]
api_key = your_anthropic_api_key

[GEMINI]
api_key = your_gemini_api_key

[OPENAI]
api_key = your_openai_api_key
model = gpt-4  # Optional: specify a default OpenAI model

[DEEPSEEK]
api_key = your_deepseek_api_key
model = deepseek-coder  # Optional: specify a DeepSeek model
base_url = https://api.deepseek.com  # Optional: custom base URL

[BEDROCK]
# AWS credentials are typically set up in ~/.aws/credentials
# You can optionally specify a profile here:
# profile = your_aws_profile

[OLLAMA]
model = llama2  # Optional: specify a default Ollama model

[GROQ]
api_key = your_groq_api_key
model = llama3-8b-8192  # Optional: specify a Groq model

[SAMBANOVA]
api_key = your_sambanova_api_key
model = Meta-Llama-3.1-405B-Instruct  # Optional: specify a SambaNova model

[OPENROUTER]
api_key = your_openrouter_api_key
model = openai/o1-mini-2024-09-12  # Optional: specify a default OpenRouter model

Each section corresponds to a different AI provider or model type. You only need to include the sections for the models you plan to use. The BASE section is used to set the default model for Hermes.

Replace your_*_api_key with your actual API keys for each service. For some services, you can optionally specify a default model or other settings like base URLs.

For AWS Bedrock, Hermes uses the default AWS credentials setup. If you need to use a specific AWS profile, you can uncomment and set the profile option in the [BEDROCK] section.

Contributing

We welcome contributions to Hermes! If you have ideas for improvements or have found a bug, please feel free to:

• Submit pull requests
• Create issues
• Suggest new features

Your input helps make Hermes better for everyone.

License

Hermes is licensed under the Apache License 2.0. For full details, see the LICENSE file in the project repository.