Termbot is a command-line tool for interacting with OpenAI’s GPT directly from your terminal.
ℹ️ Version Note
The original Termbot (Python) was fully rewritten in Go (Golang) as of version v2.0.0.
The previous Python version is still available here - https://github.com/Argandov/termbot/tree/v1.0.0 under the v1.0.0 tag.
- Clone the repository:
git clone https://github.com/Argandov/termbot.git- Finish setup:
- Run
go build .and move the termbot binary to your bin path
Or simply run the helper script setup.sh. Examine the setup script first and adjust to your specific system setup if necessary.
- Run termbot setup
termbot --setupIt will perform the following operations:
- Create a
$HOME/.envfile, - Will ask for an OpenAI API Key, and upon receiving it, it will add it to your previously created
.envfile, - Create termbot config folder inside
$HOME/.config/termbot, - Add a default System prompt ("Context") file in
$HOME/.config/termbot/context/default
- Adjust as needed
Modify the file $HOME/.config/termbot/context/default as needed. This is the default "system context" given to OpenAI to set the tone of the conversation. DO NOT change the name of this default file. It is the base system prompt used, unless specified otherwise (See "context" below in this README).
Add additional contexts depending on your needs or what you want to do with Termbot.
This versatile tool can be used in Interactive, Prompt, or Context modes with numerous command-line options available.
Usage:
termbot [flags]
Flags:
-c, --context string Choose a context to use (default "default")
-d, --dry-run Print the prompts without actually running them
-e, --examples Print examples and additional usage
-f, --file string File input (absolute or relative)
-h, --help help for termbot
-l, --list-contexts List available contexts
-p, --prompt string LLM prompt (required)
--setup-checklist Make termbot set itself up in your system
-m, --slim-mode Enable slim mode (No ANSI colors)
-v, --verbose Enable verbose mode
If ran from a non-TTY, need to pipe termbot's output to a file, or ANSI colors is not supported, use always with --slim-mode or -m.
Termbot can handle input from local files or piped stdout, making it quite flexible in terms of usage:
# Analyzing local files
termbot -f config.yaml -p "Explain what this file does and analyze any potential errors"
# Piping stdout into termbot
cat README.md | termbot -p "What is this file about?"Or, "zero prompt mode":
termbot -f script.py -c code_analyzerwhere code_analyzer could be a generic context or prompt, that "analyzes code" or something.
aliasing Termbot:
alias t=termbot
Also, if ANSI colors are not available (slim mode), we can simply alias -m together
In this mode, you can create or edit custom context files in the context directory. This will provide instructions on how the Termbot responds to the given prompts, the so-called personality or 'mood'. You can list the current contexts by using the command termbot -l. This idea is 100% inspired by Daniel Miessler's Fabric's usage of "patterns".
Context files are inside $HOME/.config/termbot/context/ folder. They do not need to follow any format and can be raw files without any extension (i.e. "default", "analyze_document", etc).
Requires input file -f flag and a context file -c flag:
termbot -c analyze_script -f script.py
Where "analyze_script" is a context file with specific instructions.
-f can be used with either absolute or relative paths (The program with first seek relative, then absolute path if the first fails).
Hence:
termbot -f hello.py
Will seek first for hello.py in the current directory. If it's not found, it will search at /hello.py
Dry Run (-d) will print the prompt WITHOUT sending it to the LLM. This is useful for testing purposes (Whem using stdin pipes, files, etc.)
Verbosity (-v ) will simply be explicit and print the user prompt, system prompt and the LLM response.
beta: Early prototype versionv1.0.0: Final Python-based Termbot version (view here)v2.0.0: Full rewrite in Go — this is the current and recommended version