The CodeMD blog

Real findings, not generic advice. Every post links to the advisories, commits and reports behind it so you can check the evidence yourself.

Case studies

14 real MCP server vulnerabilities, replayed through a pull request check

Published command-injection advisories, each traced to the commit that added the bug and the risk path CodeMD flagged on it.

Scan results

What CodeMD found scanning six popular open-source repos

express, gson, Humanizer, click, sinatra and cobra: what was reviewed, what the AI concluded, and why the honest result is no confirmed vulnerabilities.

Guide

How to give Claude access to your GitHub repository safely

A fine-grained token for one repo, saved in an environment variable, tested without printing it, and revoked when you're done.

Product update

CodeMD now names the feature behind every change

A real Feature Impact check: the feature is discovered automatically, with an explanation of what changed and whether any test reaches it.

Guide

How to verify AI-written code before you merge it

Five questions to ask about every AI-generated pull request.