Search code, repositories, users, issues, pull requests...

The idea of DevDocs is to ensure that software engineers and (LLM) software devs dont have to go through copious amount of tech documentation just to implement the tech.

Problem: LLMs are great but they are OUTDATED, as a software developer I had to constantly refer to product knowledge from LlamaIndex, AutoGen and other Machine Learning frameworks to build our Excalibur Platform for cybersecurity and have LLM breakdown the problem. I was copy pasting content from each page into Claude to understand the context and break it down for me and soon the LLM started hallucinating.

Solution: I built a unique platform called DevDocs which spiders through all the child pages of the primary URL, scans if the webpages are live or dead, parses every page (FREE of cost) and converts it into markdown for easy LLM digestion. You can take the markdown and embed into your vector database or use an MCP server like obsidian and load it into Cline for software development.

Goal: I wanted to use state of the art technology but wanted the speed of an up to date LLM, cant do that without an hashtag#MCP(model context protocol) server. So now if I wanted to implement a vector database into my code, I just copy the URL, paste it in DevDocs and the tool grabs all the data from its child websites and spits it into markdown which is then uploaded into my MCP server for Claude to use. This cuts down weeks of research into just a bunch of questions.

Leave a Comment