shivji2138/devmentor-ai

DevMentor AI is a command line tool that acts like a pair programmer you can talk to directly from your terminal, with one key difference from a typical AI chat assistant: it remembers things about your project over time. Instead of explaining your stack, your bugs, or your project quirks over again in every session, you can teach it once and have it recall that context in later conversations. This long term memory is built using an approach the author calls Hindsight inspired memory, stored locally in a SQLite database on your own machine, so nothing about your projects needs to leave your computer beyond the actual questions you send to the AI model. For generating answers quickly, DevMentor connects to Groq, a service known for very fast AI response times, so replies come back with low latency compared to some other providers. The command line interface itself is built with two Python libraries, Typer for structuring commands and Rich for colorful, readable terminal output. Getting started involves cloning the repository, optionally setting up a Python virtual environment, installing the listed dependencies, and creating a small configuration file that holds your own Groq API key, which you will need to obtain separately since DevMentor does not provide one for you. Once configured, you interact with it through two main commands: one to chat and ask a question, where it automatically pulls in whatever relevant memory it has stored, and another to manually tell it something specific you want it to remember for future sessions. Under the hood, the project is organized into a handful of small, focused files: an entry point that sets everything up, a file defining the actual chat and remember commands, a file that talks to the Groq API, a file that manages how memories are stored and retrieved, and a file that sets up the local database the first time you run it. The project is open source under the MIT license and welcomes outside contributions through pull requests or issues.