We ran NSPEC against its own landing page. Here’s what broke.
A few hours after the marketing site went live, we pointed a standard-tier NSPEC run at nspec.dev. Four real bugs came back · every one of them a thing I thought I had already checked.
Why the verifier is a separate agent, not a retry loop
Retry loops inherit the bias that produced the finding. A verifier is a different agent with its own context, its own budget, and the discipline to say “I could not reproduce this.” Here is how that changes the shape of the bug report.
QA that pays for itself by bug #2
A concrete ROI case for autonomous QA. The math on staging bugs vs prod incidents, the engineer-hour math, and why most QA tooling quietly loses money.
Filing tickets over MCP: Jira, Linear, GitHub
Why we dropped webhooks and CSV imports in favour of Model Context Protocol for tracker delivery. Field mapping, severity → priority, owner inference, and the bit where idempotency stops mattering.
The five categories of bugs that manual QA keeps missing
A taxonomy of the bugs our beta runs catch most often · the classes of defects that slip through code review, CI, and human QA because they’re invisible to the processes that are supposed to find them.
New posts land on the waitlist email · join on the home page.