Software Engineering :: Pair Programming :: AI Assistant :: RepoPrompt :: About :: How I use Repo Prompt to build Repo Prompt
⪼ Made with 💜 by Polyglot.
This video is a tutorial and product demo by Eric, the creator of Repo Prompt, aimed at educating developers on how to use Repo Prompt to enhance their own development workflow—specifically by showing how he used Repo Prompt to build and extend itself. Eric demonstrates how to integrate a new provider (Grok) into the app using AI-assisted workflows that automate code search, editing, and implementation using parallel model runs. The video showcases the power of combining tools like AI Studio, Gemini, and DeepSeek with Repo Prompt's code mapping and prompt orchestration capabilities.
- Automatically generates a scoped file list for a task using an AI-driven context builder
- Creates compressed code maps to make large codebases navigable by LLMs
- Allows selective file inclusion and supports parallel token-aware prompting
- Sets up a task to add Grok support by generating a file list and prompting edits
- Uses multiple models (Gemini, 03, DeepSeek) to create and apply XML-based edit instructions
- Applies edits, resolves merge conflicts, and tests compiled results in-app
- Uses XML to batch and orchestrate model edits across multiple files
- Compares outputs from different LLMs for precision vs. speed (e.g., Gemini vs. 03)
- Fixes compiler issues iteratively using GPT-4.1 and model-assisted patching
- Successfully adds and configures Grok 3 Mini model
- Demonstrates full end-to-end reasoning test inside the app
- Shows final working integration and reflects on how cost-effective and fast the process was