Videos
Tutorials, deep dives, devlogs, and commentary on AI and agentic coding — from @UltraDotDev.
Anthropic Drops Fable 5 1 And Usage Is Nuts #claudefable5 #claude #claudecode #fable5 #aicoding #ai
A quick reaction to Claude Fable 5.1 and the usage limits encountered while trying it.
Anthropic Deleted 80% of Claude’s Prompt. I Went Further. #aicoding #codex #claudecode #ai #glm53
Moving from Claude to Codex without losing project context: how Ultra Cortex uses a graph database to carry knowledge between coding agents.
Ox Alpha Unmasked! #aicoding #openweights #openrouter #claudecode #codex
Ox Alpha is revealed as GLM-5.3-Flash. A look at the model, its Ollama availability, and plans for a head-to-head build against DeepSeek V4.
How #claudecode Failed Me #aicoding #codex
Why I moved client work from Claude Code to Codex, and how trust, control, and reliable production delivery shaped my AI coding workflow.
Your AI Has No Design Taste — Here's The Fix #Shorts
A quick introduction to Design Substrate, the free skill that gives AI a design vocabulary beyond the default indigo gradient.
Your AI Has No Taste Because You Never Gave It Any
Building DesignSubstrate.com with a free AI skill: exploring design taste, surprise mode, and a complete design system from a blank prompt.
GPT 5.4 Just Did In 12 Minutes Something 5.2 Couldn't Do In Three Hours
Putting GPT-5.4 to work on a Rust terminal emulator, from the first compile to tabs, split panes, themes, and the limits of agentic coding.
I Replaced VS Code With an IDE I Built Entirely With AI
Building TIDE, a full-featured IDE, in six weeks with AI: the architecture, multi-model workflow, and features behind my VS Code replacement.
#ai #vibecoding May Be Destroying #opensource - #opensourcesoftware #aicode
How AI coding affects open source, with security findings from CodeRabbit, Veracode, and Escape and a call to review generated code before shipping.
The #vibecoding Report Card Is...Not Great! #ai #aicode #vibecode
A quick reality check on vibe coding: why working software can still contain security holes and why generated code needs careful review.
#vibecoding Is Bringing About The Rise of #vibehacking
What happens when unreviewed AI-generated code meets an attacker? The rise of vibe hacking, illustrated by the Lovable app security failure.
18,000 Users from College Student to Minors Caught in #lovable #vibecoding #vibehacking
A breakdown of the Lovable app vulnerabilities that exposed 18,000 users, plus the security checks to run before shipping AI-generated code.
Secure it ain't: #lovable + #vibehacking == 18k Users Exposed from bad #ai #vibecoding
A short look at the Lovable app security failure that exposed 18,000 users and what it reveals about shipping AI-generated code without review.
#tailwindcss Is In Deep Trouble Due To The #ai #opensource crisis #digitalmarketing
How AI-driven changes in developer habits put pressure on Tailwind CSS revenue, staffing, and the business model behind open-source tools.