Hey Product Hunt 👋 We built CodeMind AI around a frustration we kept running into: AI coding tools can sound incredibly confident even when they're wrong about your codebase. So instead of starting with “chat with your repo,” we started with evidence. CodeMind indexes your repository, parses the code with Tree-sitter, resolves imports into a real dependency graph, and grounds its answers in exact file paths and line ranges. That same principle runs through the rest of the product. 🔎 Ask questions about an unfamiliar codebase and see exactly where the answer came from. 🏗️ Explore an architecture graph built from actual dependencies rather than just folder structure. 🐛 Run targeted bug, security, and performance checks where every finding points back to the code that triggered it. 👀 Review PRs without the noise—CodeMind analyzes the lines that actually changed and anchors comments to the diff. 🛠️ When it finds something worth fixing, turn the finding into a draft PR with a fix and test for you to review. The goal is for CodeMind to feel less like a chatbot telling you “trust me” and more like a staff engineer saying: “Here’s what’s happening. Here’s the code that proves it. And here’s how I’d fix it.” We have a free plan for one repository, so you can try it on a real codebase without paying. I’d especially love feedback from the PH developer community on three things: Do the citations make you trust the answers more? Does the architecture graph match how you think about your system? And what code checks should we build next? Thanks for checking out CodeMind AI 🙌 The technical details in that comment match CodeMind's current implementation and product positioning: Tree-sitter indexing, resolved imports, subsystem architecture mapping, evidence-backed checks, changed-line PR analysis, and fix proposals.

CodeMind AI

AI staff engineer that reads your code and shows its work

AI staff engineer that reads your code and shows its work