Kel
Kel is an AI-based assistant embedded in your Command Line Interface (CLI). It is designed to augment productivity by automating repetitive...
Last verified:
What is Kel?
Kel is a CLI-based AI assistant that integrates directly into your terminal, allowing you to query large language models (LLMs) like OpenAI, Anthropic, and Ollama for quick answers and insights. It takes natural language questions, sends them to the specified LLM using your own API keys, and displays responses right in the CLI, enhancing workflow efficiency for developers and testers. Key features include support for multiple LLMs, customizable prompts, OpenAI Assistants for chatting with uploaded files, and a configurable config.toml file for styling and settings.
Designed for software engineers, QA professionals, and CLI enthusiasts who want to automate repetitive tasks and get intelligent assistance without leaving the terminal. It enables chatting with documents by uploading files to an OpenAI Assistant, analyzing content, and extracting insights on the fly. The tool is lightweight, requiring only Python 3.6+, pip, and your API keys to get started.
Kel emphasizes simplicity and customization, with options to switch models, adjust prompts like 'You are a demography expert', and style outputs. It's fully open-source, free to use, and supports both global queries and file-specific interactions, making your CLI smarter and more productive.
Kel pricing
Pricing model: Free
Kel is completely free and open-source with no paid tiers or subscriptions. Users bring their own API keys for LLMs like OpenAI or Anthropic, incurring only the provider's usage costs. Optional paid support plans available by contacting via the website; no free tier limitations as the tool itself has none.
Kel pros
- CLI-based for seamless terminal integration
- Supports OpenAI, Anthropic, and Ollama LLMs
- OpenAI Assistants for file chatting
- Bring your own API key flexibility
- Customizable prompts per query
- Configurable styling via config.toml
- Free and completely open-source
- Automates repetitive developer tasks
- Quick installation via pipx
- Switch LLMs with -c flag easily
- Chat mode with documents using -f
- Environment variable API key setup
- Lightweight with minimal dependencies
- Supports max tokens configuration
- Contributions welcome on GitHub
- Enhances productivity in testing workflows
- No account required, privacy-focused
Kel cons
- Requires own LLM API keys
- Python 3.6+ installation needed
- pipx setup for isolated install
- Manual config.toml configuration
- Limited to CLI, no GUI option
- API costs borne by user
- No built-in free LLM hosting
- File uploads only for OpenAI Assistants
- Support plans not detailed publicly
Frequently asked questions about Kel
What is Kel?
Kel is a CLI-based AI assistant that connects your questions to LLMs like OpenAI, Anthropic, and Ollama, printing responses directly in your terminal for quick insights and task automation.
How do I install Kel?
Install via pipx: python3 -m pip install --user pipx; pipx ensurepath; pipx install kel-cli. Then download config.toml with curl -O https://raw.githubusercontent.com/QAInsights/kel/main/config.toml, create ~/.kel/, and move the file there.
What LLMs does Kel support?
Kel supports OpenAI (including Assistants), Anthropic, Ollama for local models, and Google models are mentioned in some docs, with more coming soon. Switch with -c flag like -c ollama.
Do I need an API key?
Yes, bring your own API keys for OpenAI (OPENAI_API_KEY), Anthropic (ANTHROPIC_API_KEY), etc., set as environment variables. For free local use, set up Ollama.
How do I use OpenAI Assistants?
Use kel -a perfGPT -f <file_name> to summon an assistant with your file, enter chat mode to ask questions about the document, and it deletes upon exit.
Can I customize prompts?
Yes, add custom prompts with -p flag, e.g., kel 'question' -p 'You are a physicist' to tailor the LLM's response style.
Where is the config file?
Default at ~/.kel/config.toml or ~/.config/kel/config.toml, or set KEL_CONFIG_FILE env var. Edit for models, styles, max tokens, etc.
Is Kel free?
Absolutely free and open-source; no payments for the tool. Only LLM API usage costs apply based on your keys.
How to check Kel version?
Run kel -v to view version, kel -h for help, or kel -s openai to see config details for a provider.
Can I contribute to Kel?
Yes, it's open-source on GitHub at QAInsights/kel; contributions for features, bug fixes, or more LLM support are welcome.