ChatMastermind is a Python application that automates conversation with AI, stores question-answer pairs with tags, and composes relevant chat history for the next question.

The project uses the OpenAI API to generate responses and stores the data in YAML files. It also allows you to filter chat history based on tags and supports autocompletion for tags.

Official repository URL: https://kaizenkodo.no/gitea/kaizenkodo/ChatMastermind.git

Requirements

Python 3.9 or higher
openai
PyYAML
argcomplete

You can install these requirements using pip:

pip install -r requirements.txt

Installation

You can install the package with the requirements using pip:

pip install .

Usage

The cmm script has global options, a list of commands, and options per command:

cmm [global options] command [command options]

Global Options

-C, --config: Config file name (defaults to .config.yaml).

Command Options

Question

The question command is used to ask, create, and process questions.

cmm question [-t OTAGS]... [-k ATAGS]... [-x XTAGS]... [-o OUTTAGS]... [-A AI_ID] [-M MODEL] [-n NUM] [-m MAX] [-T TEMP] (-a QUESTION | -c QUESTION | -r [MESSAGE ...] | -p [MESSAGE ...]) [-O] [-s FILE]... [-S FILE]...

-t, --or-tags OTAGS: List of tags (one must match)
-k, --and-tags ATAGS: List of tags (all must match)
-x, --exclude-tags XTAGS: List of tags to exclude
-o, --output-tags OUTTAGS: List of output tags (default: use input tags)
-A, --AI AI_ID: AI ID to use
-M, --model MODEL: Model to use
-n, --num-answers NUM: Number of answers to request
-m, --max-tokens MAX: Max. number of tokens
-T, --temperature TEMP: Temperature value
-a, --ask QUESTION: Ask a question
-c, --create QUESTION: Create a question
-r, --repeat [MESSAGE ...]: Repeat a question
-p, --process [MESSAGE ...]: Process existing questions
-O, --overwrite: Overwrite existing messages when repeating them
-s, --source-text FILE: Add content of a file to the query
-S, --source-code FILE: Add source code file content to the chat history
-l, --location {cache,db,all}: Use given location when building the chat history (default: 'db')
-g, --glob GLOB: Filter message files using the given glob pattern

Hist

The hist command is used to print and manage the chat history.

cmm hist [--print | --convert FORMAT] [-t OTAGS]... [-k ATAGS]... [-x XTAGS]... [-w] [-W] [-S] [-A SUBSTRING] [-Q SUBSTRING]

-p, --print: Print the DB chat history
-c, --convert FORMAT: Convert all messages to the given format
-t, --or-tags OTAGS: List of tags (one must match)
-k, --and-tags ATAGS: List of tags (all must match)
-x, --exclude-tags XTAGS: List of tags to exclude
-w, --with-metadata: Print chat history with metadata (tags, filenames, AI, etc.)
-S, --source-code-only: Only print embedded source code
-A, --answer SUBSTRING: Filter for answer substring
-Q, --question SUBSTRING: Filter for question substring
-l, --location {cache,db,all}: Use given location when building the chat history (default: 'db')
-g, --glob GLOB: Filter message files using the given glob pattern

Config

The config command is used to manage the configuration.

cmm config (-l | -m | -c FILE)

-l, --list-models: List all available models
-m, --print-model: Print the currently configured model
-c, --create FILE: Create config with default settings in the given file

Print

The print command is used to print message files.

cmm print (-f FILE | -l) [-q | -a | -S]

-f, --file FILE: Print given file
-l, --latest: Print latest message
-q, --question: Only print the question
-a, --answer: Only print the answer
-S, --only-source-code: Only print embedded source code

Examples

Ask a question:

cmm question -a "What is the meaning of life?" -t philosophy -x religion

Display the chat history:

cmm hist

Filter chat history by tags:

cmm hist --or-tags tag1 tag2

Exclude chat history by tags:

cmm hist --exclude-tags tag3 tag4

List all tags and their frequency:

cmm tags -l

Print the contents of a file:

cmm print -f example.yaml

Configuration

The default configuration filename is .config.yaml (it is searched in the current working directory). Use the command cmm config --create <FILENAME> to create a default configuration:

cache: .
db: ./db/
ais:
  myopenai:
    name: openai
    model: gpt-3.5-turbo-16k
    api_key: 0123456789
    temperature: 1.0
    max_tokens: 4000
    top_p: 1.0
    frequency_penalty: 0.0
    presence_penalty: 0.0
    system: You are an assistant

Each AI has its own section and the name of that section is called the 'AI ID' (in the example above it is myopenai). The AI ID can be any string, as long as it's unique within the ais section. The AI ID is used for all commands that support the AI parameter and it's also stored within each message file.

Autocompletion

To activate autocompletion for tags, add the following line to your shell's configuration file (e.g., .bashrc, .zshrc, or .profile):

eval "$(register-python-argcomplete cmm)"

After adding this line, restart your shell or run source <your-shell-config-file> to enable autocompletion for the cmm script.

Contributing

Enable commit hooks

pip install pre-commit
pre-commit install

Execute tests before opening a PR

pytest

Consider using `pyenv` / `pyenv-virtualenv`

Short installation instructions:

install pyenv:

cd ~
git clone https://github.com/pyenv/pyenv .pyenv
cd ~/.pyenv && src/configure && make -C src

make sure that ~/.pyenv/shims and ~/.pyenv/bin are the first entries in your PATH, e.g., by setting it in ~/.bashrc
add the following to your ~/.bashrc (after setting PATH): eval "$(pyenv init -)"
create a new terminal or source the changes (e.g., source ~/.bashrc)
install virtualenv

git clone https://github.com/pyenv/pyenv-virtualenv.git $(pyenv root)/plugins/pyenv-virtualenv

add the following to your ~/.bashrc (after the commands above): eval "$(pyenv virtualenv-init -)
create a new terminal or source the changes (e.g., source ~/.bashrc)
go back to the ChatMasterMind repo and create a virtual environment with the latest Python, e.g., 3.11.4:

cd <CMM_REPO_PATH>
pyenv install 3.11.4
pyenv virtualenv 3.11.4 py311
pyenv activate py311

see also the official pyenv documentation

License

This project is licensed under the terms of the WTFPL License.

README.md

ChatMastermind