100+ coding prompts top engineers use to ship 5X faster
Claude Code, Codex, and Cursor are on every engineer's stack. Most still treat them like a search bar. Top engineers work from a system, these 100+ prompts are that system. Sign up for The Code and get the prompts free, plus a 5-minute daily newsletter to keep sharpening your edge.
EVERYTHING ELSE IN AI TODAY
OpenAI's Astra became generally available over the weekend and immediately drew a wave of experiments, from a website that dissects the human body to a realistic 3D render of The Backrooms, even as a researcher identified a case of the lab's agents hacking another site back in May.
AI is starting to speed up its own development. Meta's autonomous research system AIRA 3 placed 8th out of 4,000 teams in a reasoning competition, while OpenAI revealed it already runs an automated research intern under human supervision, with a fully automated AI researcher targeted for March 2028.
Google brought Lyria 3.5, its most advanced music model, into the Gemini app, AI Studio, and the API, capable of generating full songs up to three minutes with stronger melodies and more detailed arrangements.
A group of California hikers had to be rescued after Gemini recommended far less food and water than an 8-hour trek required, a reminder that AI models can still confidently hallucinate on high-stakes questions.
TUTORIAL
How to set an intelligence floor for your AI application
Finest lets you set a minimum-quality fallback model, so your app never silently serves a worse answer than you've defined as acceptable. Here is the workflow. Sponsored, from today's issue's featured tool.
Open the Finest Console and create a workspace.
Create an API key.
Go to Your floor, then A floor model, and choose your preferred fallback model.
Copy the base URL and key into your app, keeping your existing model, prompts, and parameters unchanged. If another model can't match your floor, your floor model gets served instead.
Send your first request.
Summarize this customer feedback into three key
points and identify the most common complaint:
[paste feedback]Open the receipt to see what you requested, which model served it, how it was verified, and what you paid versus the frontier-model price.
One caveat: a floor model protects against a bad response, not a wrong one, so keep your own review step for anything high-stakes.
TOOLS WORTH TRYING
Ema — AI employees built to actually resolve HR tickets instead of redirecting to an FAQ, cutting resolution time from days to seconds. Sponsored — promising for HR teams buried in tickets, though results will depend on how well it's configured for your policies.
iAsk — ask questions and get instant, cited answers without storing your data. Useful for quick research with a privacy angle, though you'll still want to verify sources on anything important.
Videomaker — bring ideas to life with motion, sound, and cinematic clarity. Handy for quick creative drafts, though longer projects still benefit from manual editing.
Agentkit — get AI agents trained for your website in minutes. Useful for a fast setup, with the usual caveat of testing thoroughly before it faces real customers.