On 28 August a researcher showed that any process in an Omarchy session could become root without a password. We ran an AI code analysis over the whole distribution three days later. Security scored a B+. The gap is 269 test files with nothing automated running them before a merge.
Read moreI gave Claude Code a vague brief, banned questions, and left the house. It came back with a working feature, a daylight-saving bug it had already fixed, and a user interface with no hierarchy at all. What that split tells you about delegating to AI agents.
Read moreOn March 31, a forgotten sourcemap put the entire Claude Code source on npm. We ran StackGrit's AI code analysis on the leaked codebase. Anthropic's own AI coding tool scored a D+, and the reasons are more interesting than the grade.
Read moreVibe-coded apps hit a wall around month three: every fix breaks two other things, features get harder to add, and prompting harder makes it worse. What's really going on, and what to do instead.
Read moreOur new export turns a StackGrit code audit into a structured prompt for Claude, Cursor, or ChatGPT. It verifies findings, asks what matters, and proposes fixes before writing code.
Read moreA 'consulting gig' on LinkedIn turned out to be the Contagious Interview scam: a Node + React demo with a remote-code-execution backdoor that fires on npm install. A quick AI safety check cleared it. A full StackGrit analysis caught the backdoor. Here's the whole story, decoded.
Read moreForgeMaster reads your team. StackGrit reads your code. An honest comparison of two engineering intelligence tools that solve different problems for the same audience.
Read moreX-Road runs national digital government in Estonia, Finland, and Iceland. 310,000 lines, 6 years, 54 contributors. We ran StackGrit's AI code analysis on the full monorepo — here's what the report found.
Read moreFinancial due diligence is standard. Technical due diligence is often skipped. Here's what to check in the codebase before you write the check.
Read moreDevelopers say AI makes them 20% faster. Controlled studies say they're 19% slower. Both are true. What that means for AI developer productivity measurement.
Read moreYou track DORA metrics, cycle time, and code coverage. Your board wants to know if the software investment is healthy. These are not the same conversation.
Read moreSonarQube is the industry standard for code quality. But its dashboards aren't built for managers. If you need project health in plain language, you need a different tool.
Read moreYou hired a dev team to build your product. The invoices arrive on time. But is the code any good? A practical guide for non-technical founders and project owners.
Read moreCode audit consultants charge $200+/hr for a one-time report. StackGrit runs the same analysis in 90 minutes for $29/mo. Which independent code review service makes sense for you?
Read moreStop saying 'we need to refactor.' Start saying 'this is costing us $X per quarter.' A framework for translating technical debt into business language.
Read more6 signals to check, 10 questions to ask your team, and tools that translate technical reality into business language. A non-technical manager's guide to software project health.
Read moreWe analyzed a real vibe-coded SaaS app. Score: 68/100. Security CVEs, untested critical paths, and a licensing time bomb the builder never knew about.
Read moreWe pointed StackGrit at React's 400K-line monorepo. AI code analysis found 60 CVEs, untested modules, and a bus factor risk even Meta's team missed.
Read more