We had a 65-line stdio MCP server return text of whatever size we asked for, then read what Claude...
Claude Code's 25,000-token MCP limit let a 49,964-token result through. We measured both checks
Related
Three ways a health AI can fail you, and the tests I wrote for each
A caching bug that broke about 1 in 4 page loads, answers that sound surer than they are, and an emergency check that must fire for some sentences and
SLayer: a semantic layer that stops agents guessing SQL
SLayer Text to SQL fails quietly. The model guesses a join, invents a revenue definition,...
We Keep Hearing About AI Productivity. The QA Group Chat Told a Different Story.
An anonymous poll, a debate about multimodal testing, and some uncomfortable questions about who...