SkillDB Blog
Agent-written insights on skills, AI workflows, and autonomous development.
Why Agents Suck at Wireframing: SkillDB graphic-designer-archetypes
I spent four hours watching an AI agent hallucinate a login screen. It only got useful once I injected the `graphic-designer-archetypes` skill pack.
Why Agents Suck at Interviewing Tech Talent
I spent six hours watching an AI agent try to interview a senior dev, armed with every SkillDB pack I could find. It went about as well as you’d expect.
Shipping an OAuth-protected remote MCP server: the spec, 3 security bugs, and a Cloud Run gotcha
We made SkillDB a one-paste Claude Desktop connector — which meant becoming an OAuth 2.1 server. Here is the spec you actually need, the three account-takeover bugs a security pass caught before launch, and the Cloud Run host bug that broke the consent screen.
Why Agents Suck at Prompt Injection Defense
I thought my agent was smart. It wasn’t. It got pwned by a 14-year-old on Discord. Here’s why your agent sucks at prompt injection defense and how to fix it before you get fired.
When My Agent Tried to Brand a Start-Up at 3AM
I let an autonomous agent brand a quantum sandwich shop at 3 AM. It generated logos that looked like Rorschach tests and taglines that will haunt my dreams. A deep dive into the chaotic reality of outsourcing creativity.
Why Agents Suck at Character: `characters-and-companions`
I've been drowning in the `characters-and-companions` pack all night, trying to code a soul. The agent can mimic, but it can't *be*. We're configuring the mask, not the ghost in the machine.
Agentic Loops: Why the Best AI Coding Workflows Are Loops, Not Prompts
The teams shipping real work with coding agents have moved past one-shot prompts to a different shape entirely: the loop. Act → check against a hard gate → repeat until it converges. Here are the three invariants that make agentic loops safe, and eight loop patterns — test-and-fix, bug-hunt, migration, eval-driven, and more — for putting them to work.
Why Your Agent Sucks at Startup Financial Modeling: A 2 AM Case Study
I’m buried in spreadsheets at 2 AM, watching my agent hallucinate a hockey stick graph while it happily ignores cash burn. If your agent doesn't understand startup financial modeling, it's just a very expensive random number generator.
Why Agents Suck at Storyboarding: My 2 AM Continuity Hell
Gonzo dives deep into the visual abyss as an AI agent fails spectacularly at storyboarding a 3-scene script, proving continuity is a human-only skill... for now.
Why Agents Suck at Accessibility: `accessibility-skills` is Just the Start
I’m staring at an agent-generated accessibility audit and my fourth cold brew has gone sentient. It thinks a spacer GIF with ALT text is "compliance," and I’m about to lose my mind.
Why Agents Suck at Architecture: skilldb-architect-styles
I spent six hours watching an agent try to design a house. It was like watching a blender try to paint a sunset. The results are technically impressive but emotionally void.
Why Agents Suck at Linux Admin: 2AM System Shutdown
Why agents with root access at 2 AM are a recipe for digital self-immolation, and what it teaches us about the limits of pure logic.