docs assistant

docs-assistant

Chat bot assistant for asking questions about documentation databases using LangChain.

This uses the RetrievalQA chain, in combination with the Chroma and OpenAI classes.

Usage

Steps:

1. Run the scripts in docs_assistant/loaders to load documents into the vector DB (Chroma)
2. Run any of the scripts in docs_assistant/bot to start a bot that queries the DB
3. Start chatting with the bot to query the DB and get LLM-generated responses to the question.

All scripts require the the following environment variables to be set:

OPENAI_API_KEY="..."
CHROMA_DB_DIR="/path/to/chroma/db"

Loading documents (docs_assistant/loaders)

Text files:

python -m docs_assistant.loaders.textfiles /path/to/chroma/db/ file1.txt file2.txt ...

Confluence:

CONFLUENCE_API_KEY="..."
CONFLUENCE_URL="https://domain.com/wiki"
CONFLUENCE_USERNAME="[email protected]"
python -m docs_assistant.loaders.confluence "type=page and space=SPACE"

Running the bot (docs_assistant/bot)

Within the terminal:

python -m docs_assistant.bot.cli

As a slack bot:

SLACK_APP_TOKEN="..."
SLACK_BOT_TOKEN="..."
python -m docs_assistant.bot.slack