Meta Analysis: AI Coding
Findings from measuring, auditing, and synthesizing what the tools actually do.
Category
Meta Analysis
Tags
9 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.
Filters2
Category
Meta Analysis
Tags
9 more tagsFewer tags
12 posts
Cursor is shrinking while the terminal eats the IDE
Cursor shrinking is the 18 to 12 percent work-usage drop while CLIs rose. Keep VS Code or JetBrains. Put the agent in a CLI you already run this quarter.
Claude Code is the default coding agent at work
Claude Code is the default coding agent at work in the JetBrains 2026 survey. 39% use it, 31% live in it. Check conversion, then keep a second agent ready.
OpenCode vs Claude Code on the Same Tasks: Same Passes, Different Paths
OpenCode vs Claude Code on four graded tasks. Both hit 4/4. Paths diverged on tools, wall time, and reported tokens. Product lane, not a same-model lab.
Fat Instruction Files Hurt Frontier Coding Agents
A fat instruction file costs more and can follow less. Thin rules passed every check. Fat files dropped to 80% and burned 21x the prompt tokens on the same task.
Independent verification of Bun memory usage claims
Bun's Rust rewrite claims lower memory usage. A firsthand build-loop test on the exact version Bun benchmarked checks whether those numbers actually hold up.
Claude Code vs Codex Limits, Measured in the Same Unit
Claude Code vs Codex limits, measured across 28 days and 56.8 billion tokens. Why token counts cannot predict quota burn, and what to read on each side instead.
Token-Compression Skills Do Not Survive Measurement
Token compression skills report savings the invoice never shows. Cache reads carry 87 percent of an agent bill, and cutting tokens can push the cost up.
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.