Hi everyone! 👋
I've been building a small open-source Claude Code plugin called feature-recon, and I'd love to get some feedback from people who spend their days inside large codebases.
The idea is simple:
Instead of asking an AI "is my project production ready?", feature-recon inspects the codebase thoroughly, and challenges every claim and identifies bugs, gaps, opportunities, etc.
It stores its findings as JSON files in docs/recon/* and It can help answer questions like:
- Where is this feature implemented?
- What's broken or incomplete?
- Which edge cases are handled?
- What isn't tested?
- Which files are involved?
- Where are the TODOs or technical debt?
The goal is to make AI-assisted development more evidence-based instead of guess-based.
I originally built it because I kept onboarding into unfamiliar projects and spending way too much time manually tracing features across dozens of files.
I'm actively looking for ideas, criticism, and feature requests from people who use Claude Code daily.
GitHub:
https://github.com/iSerter/claude-feature-recon
If you've ever wished AI would show its work instead of just making claims, I'd love to hear what you'd want from a tool like this.
Example Report Summary:
There are more example outputs in https://github.com/iSerter/claude-feature-recon/tree/main/example-reports
Top comments (0)