Chat on the command line using the codeveloper engine with the chatgpt command line tool

As an alternative to running the CoDeveloper GPT Engine from ChatGPT you can run it with the chatgpt script from my ChatGPT Toolsuite. This not a full fledged chat interface, (hence the script name pmcodevgpt= “Poor Mans CoDEVeloper GPT”), but starts up within a second and can be used for quick tasks. You'll need an OpenAI API key to use it, though - either in an environment variable OPENAI_API_KEY or in a file ~/.openai-api-key.txt . The pmcodevgpt script starts the CoDeveloper GPT Engine in the background and the chatgpt script with a tools definition that is generated from the OpenAPI description of the CoDeveloper GPT Engine.

If you call pmcodevgpt, you can type your prompt and end it with /end on a line of its own, or press Ctrl-D to end each message. After processing and printing the response this starts again - abort the program with Ctrl-C.

If you like you could also use the audio chat feature of the chatgpt script to talk to the CoDeveloper GPT Engine. You can call pmcodevgpt -ca to start the audio chat - follow the instructions it prints to the console. You can dictate your prompts, but the output will be written to the console.

BTW: if you know any open source models / interfaces that support a function calling / tools interface like OpenAI does, please let me know! I'd like to try that / integrate that, too.