I built DocComment to solve the challenge of understanding unfamiliar code quickly, whether it's legacy code, AI-generated, or poorly documented.

The key technical difference from existing tools like Copilot's commenting feature is that DocComment analyzes code structure before generating explanations. It builds a structural representation of both the specific code snippet and its broader context, allowing the LLM to generate more precise and contextual documentation. It's not competing against Copilot or Cursor, but rather work with them.

Technical details:

1. Operates alongside code without modifying source files

2. Uses structural analysis to determine appropriate detail level for comments

3. Focuses on explaining both local code behavior and broader business context

4. Integrates with git repos for full codebase context(Planning)

Current results show improved accuracy compared to pure LLM-based commenting systems, particularly for:

1. Large functions with complex logic

2. Code with unclear variable names

3. AI-generated code

4. Business logic heavy sections

Would love feedback from the HN community. Both business and techincal perspectives are welcomed.