botman lets you talk to an LLM. It is optimized for use in the terminal. You can use stdin or an argument for input, outputs content to stdout and errors to stderr.

Note that botman does not run any commands. It does not have the ability to do anything, as having some automated LLM control your terminal couldn't possibly lead to anything good. So botman simply shows you the output.

You can pipe the output of the last received response to any other shell command, e.g. botman -l | bash. Use this at your own risk.

Supported LLMs

Currently botman can use OpenAi, Claude and FireworksAi. This includes:

GPT 4o
Claude 3.5 sonnet
Llama 3 8B / 70B
Mixtral 7B / 22B

Here is a complete list all supported models.

Install from source

Clone the repo
Run go install .
Run botman --init to setup the config.
(optional) Create an alias in your shell. e.g. echo 'alias bot="botman"' >> ~/.bashrc
Test that it works by running botman "say hi" or bot "say hi"

Examples

# Open a new interactive chat.
botman

# Use an argument as input.
botman "tell me a joke about the golang gopher"
botman "git command for undoing last commit"
botman "untar .tar.gz file into new directory"

# Use stdin as input
echo Quote a Bob Kelso joke | botman

# Use both
ls -al | botman "Which files are hidden?"
cat deployment.yaml | botman "how many replicas will this run?"

# Print the last received response
botman -l

# Pipe the last response into bash (at your own risk)
botman -l | bash

# Pipe the last response into a file
botman -l >> some-file.go

# Show the last conversation
botman --history 0

# Show the next-to-last conversation
botman --history 1

# Change model and/or set API keys
botman --init

demo

Interactive mode

In interactive mode, the program does not exit after a response, so you can continue the conversation.

By not supplying any arguments or stdin, botman will run in interactive mode.

Use interactive mode explicitly combined with stdin or arguments by giving the -i flag.

# Start interactive mode without an initial prompt
botman

# Start interactive mode with arguments
botman -i "How many bees in a bonnet?"

Data privacy

botman talks directly to the OpenAi API. So assume that OpenAi knows about your plans to overthrow goverments and such. Other than that, botman does not reach out to any service. It does store your chat history locally in ~/.botman/history. You can disable this in the settings file ~/.botman/config.yaml by setting saveHistory to false.

Motivation

I created it mainly for myself but thought it might be useful for others. My motivation stems from seeing some closed-source CLI-LLM integrations from companies I don't necessarily trust. So, I created something free and open source for those of us who value open source and transparency. (Yes, it does still use OpenAI's API, but I am working towards abstracting that away so it could use any LLM.)

Roadmap

I'm adding features as I feel they're useful. I'm open to suggestions and outside contributions. The aim is to be simple, non-intrusive and transparent about data.

# Packages

# README