DynaDocs

• 

  homescreen

• 

  file preview

• 

  chatbot

• 

  hashing

Inspiration

Inspired by Mintlify's simple to use docs and our own personal need for better documentation over fast growing codebases (which AI has trouble grasping)

What it does

It is a CLI that runs within the directory of your project; it will automatically generate documentation (summaries, file structure, etc) across your entire codebase. It will also generate correctly formatted docstrings across all your methods; this will give AI that reads your codebase a better understanding of what your code does and what it's used for. We also built a codebase chatbot with Letta that allows you to query your codebase and gain insights from the documentation/docstrings we generate. Currently we support Javascript, Typescript, and Python.

How we built it

We used Term.Ink, a React based library for building CLI applications (used to build Claude Code and the new OpenAI Codex!), Google Gemini API, TreeSitter for code parsing, and Letta for the codebase aware chat.

Challenges we ran into

• learning how to use Treesitter to extract important information from codebases
• Utilizing Letta's agents with their embeddings to build the chatbot
• Learning Term.Ink, a new library ## Accomplishments that we're proud of -Clean Interface
• Blazing performance because of carefully chosen optimization ## What we learned
• AI is really good at coding... but it could be better
• Devtools are awesome!! ## What's next for DynaDocs
• Integrate this into other tools that we're building that involve deep understanding of complex codebases

Built With

• gemini
• google
• letta
• term.ink
• typescript