llm-cli

Command-line interface for querying large language models

table of contents

• install
• usage
• api
• contributing
• license

install

pip install llm-cli

usage

# basic query
llm-cli "explain recursion in python"

# specify model
llm-cli --model gpt-4 "write a haiku about debugging"

# streaming output
llm-cli --stream "tell me a story"

# read from stdin
cat input.txt | llm-cli --stdin

# save to file
llm-cli "generate test data" > output.json

# set temperature
llm-cli --temperature 0.2 "what is 2+2"

# multi-line input
llm-cli --multiline
> first line
> second line
> ^D

api

command-line options

flag	description	default
--model, -m	model to use	gpt-3.5-turbo
--temperature, -t	sampling temperature (0.0-2.0)	0.7
--max-tokens	maximum response length	2048
--stream, -s	stream output token by token	false
--stdin	read prompt from stdin	false
--multiline	enable multi-line input mode	false
--system	system prompt	null
--api-key	api key (or use env var)	$LLM_API_KEY

environment variables

export LLM_API_KEY="your-api-key"
export LLM_DEFAULT_MODEL="gpt-4"
export LLM_API_BASE="https://api.openai.com/v1"

configuration file

Create ~/.llm-cli/config.json:

{
  "default_model": "gpt-3.5-turbo",
  "temperature": 0.7,
  "max_tokens": 2048,
  "api_key": "your-api-key"
}

supported models

• OpenAI: gpt-4, gpt-3.5-turbo
• Anthropic: claude-3-opus, claude-3-sonnet
• Local: ollama/llama2, ollama/mistral

contributing

prs welcome. open an issue first for big changes.

license

MIT