Meta Analysis
Findings from measuring, auditing, and synthesizing what the tools actually do.
Category
Meta Analysis
Tags
18 more tagsFewer tags
Newsletter
New posts land in your inbox when they publish. No spam, unsubscribe anytime.
You're in. New posts head your way.
Filters1
Category
Meta Analysis
Tags
18 more tagsFewer tags
17 posts
Vercel's scriptc Compiles TypeScript to Native Binaries Without a Runtime
scriptc compiles TypeScript to 200KB native binaries with no JavaScript engine. What works, what falls back to quickjs-ng, and what npm deps mean for you.
Every Model Release Ages Your Skills
Claude Code ignoring CLAUDE.md is often an aged file, not a model bug. See which instructions a release expires, and which lines to delete on day one.
Perplexity Isn't Shutting Down. Its Citation Corpus Might Shrink
No, Perplexity is not down. The lawsuits threaten which sources it can read, so plan for a thinner citation set inside a healthy 200, and detect it in code.
What Opus 5's Novel-Level Reasoning Benchmark Actually Means
Opus 5 novel reasoning rests on one number, 30.2% on ARC-AGI-3, and it scores action efficiency, not problems solved. Here are the four config lines to change.
WebMCP Has 20× the Interest and Zero Deployments
WebMCP adoption is zero across 111,076 scanned domains while search interest holds near 20 times its January level. Here is what to ship now, and what to watch.
6 Critics Grade My Post, Then Fix the Skill That Wrote It
Six critics grade the draft, then patch the SKILL.md that wrote it. The merge code, verdict ladder, and guardrails behind a self-improving Claude Code skill.
Docker Security Audit: What I Actually Found
I ran a security audit on my self-hosted Docker stack and found 6 real issues that every hardening guide missed. Covers network exposure, secrets in env vars, base image risks, and automated scanning with Trivy and Dockle.
AI Agents in Production: What Actually Works After 6 Months
Most AI agent projects die between demo and production. After 6 months running agents at scale, these are the patterns that survived: model routing, trace-based debugging, loop detection, and hard cost ceilings.