chainchat 0.1a1

Chat with LangChain LLM models, with tool calling and multimodal support

ChainChat - Chat with LangChain

CLI to chat with any LangChain model, also supports tool calling and multimodality.

Chat Models

ChainChat will introspect any installed langchain_* packages and make any BaseChatModel subclasses available as commands with the models attributes as options - chainchat <model-command> --<option> <value>.

Just pip install any model packages you want to use, and they will be available as model commands to chat with:

$ chainchat chat
...
Commands:
  preset  Load a preset model from YAML.
$ pip install langchain_openai langchain_anthropic
$ chainchat chat
...
Commands:
  preset                          Load a preset model from YAML.
  anthropic                       See...
  anthropic-messages              See...
  azure-open-ai                   See...
...

API Keys

API keys are accessed via environment variables. By default they are loaded from a .env file located in the current directory. You can specify a different file using the chainchat --dotenv option.

OpenAI Compatible Models

You can use any OpenAI compatible model with ChainChat. For example to use xAI Grok put your XAI_API_KEY in your .env file and alias it to OPENAI_API_KEY:

$ chainchat --alias-env OPENAI_API_KEY XAI_API_KEY chat --tool read_file --prompt "Read and summarize the file ./LICENSE.txt" open-ai --model-name grok-beta --openai-api-base https://api.x.ai/v1
I am reading the file ./LICENSE.txt to summarize its contents.
...