Launched this week

KodHau: Senior Context for AI Agents
Your AI agent doesn't know what your senior engineer knew.
12 followers
Your AI agent doesn't know what your senior engineer knew.
12 followers
Before your agent writes a single line of code, KodHau MCP injects the tribal knowledge of YOUR dev team β architecture decisions, constraints, rejected approaches, and review comments that were never documented. We tested on dotnet/runtime repo, one of the most complex repos. Cursor alone said "yes, you can delete it." Cursor with KodHau surfaced 3 PRs explaining exactly why that would break prod. One tool. Runs locally. Your code never leaves your machine.


Hey Product Hunt! π
I'm Zhasulan, 17 y.o founder from Astana, Kazakhstan. I'm the builder of KodHau.
We are all vibe coding now. It lets you iterate and ship fast, but itβs dangerous. Why? Because AI agents are brilliant in writing code that COMPILES. And sometimes, compiling doesn't mean that your code is fine. AI agents like Claude Code or Cursor read your current code, but they have zero access to the Tribal Knowledge β the why behind the code that lives in your closed PRs, architectural review comments, and "don't-touch-this" warnings written by your Senior developer.
The Problem: Cursor or Claude will often suggest a "cleaner" way to write a function, unaware that a Senior Engineer rejected that exact approach 2 years ago because of a specific architectural constraint or a design decision in production.
The Solution: KodHau-MCP. It's an MCP server that acts as the Senior Engineer who knows all architectural decisions in your repo for your agent. Before AI writes code, it analyzes the history of discussions from GitHub PRs related to that specific file and function.
The Proof: We tested KodHau MCP on the github.com/dotnet/runtime repo (one of the most complex codebases on Earth, with over 50k PRs and 70k issues).
On our demo, you can see how:
Without KodHau: Cursor suggested deleting a specific block of code to "optimize" it.
With KodHau: The agent instantly surfaced 3 relevant PRs from the Microsoft team explaining that deleting this block would break the JIT compiler for specific architectures.
Why we are here: We are applying for YC Application with Product Hunt today because we believe the next era of development isn't about "better models," it's about Better Context.
I'm here all day to answer technical questions about MCP, GitHub API limits, or how we handle tribal knowledge injection to AI Agent, when they need it the most.
Vibe coding is fast. KodHau makes it safe π