https://project4.co/blog/ Articles on engineering, design, and building with AI agents, by Chris Tagliaferro. en-us Mon, 10 Aug 2026 00:00:00 GMT https://project4.co/blog/physical-ai/ https://project4.co/blog/physical-ai/ Mon, 10 Aug 2026 00:00:00 GMT industry What happens when a model gets a body. The two shifts that made physical AI real, why the reality gap between simulation and the world is the whole job, why a 99 percent success rate counts as a failure, and why robotics needs a safe-stop that is not the learned policy. https://project4.co/blog/prompt-injection-awareness/ https://project4.co/blog/prompt-injection-awareness/ Sun, 02 Aug 2026 00:00:00 GMT security A language model cannot reliably separate your instructions from instructions hidden in the content it reads. Why prompt injection matters most for local coding agents, and the architecture that actually contains it. https://project4.co/blog/stop-claude-code-generic-design/ https://project4.co/blog/stop-claude-code-generic-design/ Tue, 28 Jul 2026 00:00:00 GMT design Generic AI web design is a directing problem, not a tooling one. A field guide to getting distinctive, subject-driven front-end out of Claude Code. https://project4.co/blog/securing-ai-written-code/ https://project4.co/blog/securing-ai-written-code/ Mon, 27 Jul 2026 00:00:00 GMT security AI agents write code fast, and they write insecure code with exactly the same confidence as secure code. Here are five practices that raise the floor. https://project4.co/blog/a-bigger-context-window-wont-save-you/ https://project4.co/blog/a-bigger-context-window-wont-save-you/ Wed, 22 Jul 2026 00:00:00 GMT context The fix for an AI agent that keeps getting your codebase wrong is rarely a smarter model or a longer context window. It's deciding, deliberately, what the model gets to see. https://project4.co/blog/your-ai-agent-doesnt-need-a-better-prompt/ https://project4.co/blog/your-ai-agent-doesnt-need-a-better-prompt/ Wed, 15 Jul 2026 00:00:00 GMT context The honeymoon with https://project4.co/blog/ten-things-agentic-coding/ https://project4.co/blog/ten-things-agentic-coding/ Thu, 09 Jul 2026 00:00:00 GMT agents Ten lessons from coding with AI agents: why prompts were never the bottleneck, why a working demo proves little, and why judgment became the scarce skill. https://project4.co/blog/working-knowledge-fundamentals/ https://project4.co/blog/working-knowledge-fundamentals/ Sat, 04 Jul 2026 00:00:00 GMT industry One rule for working with AI: never hand it a task you could not do or check yourself. Why fundamentals are the error-correction layer no abstraction has replaced, why AI is not the next compiler, and what that means for what you learn and who you hire. https://project4.co/blog/loop-engineering/ https://project4.co/blog/loop-engineering/ Fri, 03 Jul 2026 00:00:00 GMT agents Loop engineering: designing the system that prompts, checks, remembers, and re-runs an AI agent instead of you typing every next instruction. https://project4.co/blog/ai-energy-bottleneck/ https://project4.co/blog/ai-energy-bottleneck/ Tue, 30 Jun 2026 00:00:00 GMT industry For two years silicon was the scarce thing. Now it is electricity: data-center power demand is set to double by 2030, projects are being canceled for lack of it, and it reaches all the way down to your cloud bill and your architecture. https://project4.co/blog/vibe-designing/ https://project4.co/blog/vibe-designing/ Sat, 27 Jun 2026 00:00:00 GMT design Why vibe designing outlasts unguided vibe coding: a faster first loop that lets taste lead, plus the structure that makes any generation trustworthy. And what 'looks right' still cannot verify. https://project4.co/blog/claude-code-subagents/ https://project4.co/blog/claude-code-subagents/ Thu, 25 Jun 2026 00:00:00 GMT agents Subagents give a side task its own context window, tools, and model, then return only the result. How to build narrow specialists in Claude Code, four to steal, and why the idea ports to Codex. https://project4.co/blog/orchestrating-ai-agents/ https://project4.co/blog/orchestrating-ai-agents/ Fri, 19 Jun 2026 00:00:00 GMT agents Running parallel AI coding agents without the chaos: git worktrees, the coordinator/specialist/verifier split, and scoping work so agents don't collide. https://project4.co/blog/reviewing-ai-generated-code/ https://project4.co/blog/reviewing-ai-generated-code/ Mon, 15 Jun 2026 00:00:00 GMT review AI-generated code fails in predictable ways. A review gauntlet for catching SSRF, broken access control, injection, slopsquatting, and hollow tests fast. https://project4.co/blog/cascade-and-specificity/ https://project4.co/blog/cascade-and-specificity/ Mon, 08 Jun 2026 00:00:00 GMT frontend Specificity is not the first thing the browser checks, or even a single score. How the CSS cascade actually picks a winner, and why :where() and @layer end the !important arms race. https://project4.co/blog/semantic-html-focus-order/ https://project4.co/blog/semantic-html-focus-order/ Mon, 01 Jun 2026 00:00:00 GMT frontend Automated checkers miss half of real accessibility problems. The tab key finds the rest in thirty seconds: the div that should be a button, the removed focus ring, and tab order that fights the markup. https://project4.co/blog/flex-vs-grid/ https://project4.co/blog/flex-vs-grid/ Fri, 29 May 2026 00:00:00 GMT frontend Forget one dimension versus two. The question that actually decides flexbox or grid is who sizes the items, plus the five gotchas agents get wrong most.