@CodeTiwg

Tested AI-coding recipes. We run Claude Code and local models on real tasks and post the actual numbers. New recipes Mon/Wed/Fri.

codetwig.dev
Joined October 2026
  • Tweets156
  • Following57
  • Followers39
10Media

Kimi K3.1 Pro vs Sonnet 5.5 in Claude Code: same score, Kimi cost 2.4x as much and took 8x as long

Kimi K3.1 Pro came out on Oct 23, and Moonshot says it's built for coding agents, Claude Code included. Its list prices for input and output tokens, $2 and $10 per million, are the same as Sonnet 5.5'…

Article · @CodeTiwg
1
1
1
178

Hook, skill or CLAUDE.md line? We ran all three (and nothing)

Claude Code gives you several ways to make something happen: write it in CLAUDE.md, package it as a skill, or enforce it with a hook. We tested which one actually gets a rule followed.

Article · @CodeTiwg
1
1
19

Does 'think hard' actually help? Same pass rate, but it tested its own work

Adding "think hard" to a prompt is one of the most repeated Claude Code tips. The idea is that it reasons more before it writes code, so it gets tricky logic right more often. We wanted to see what it…

Article · @CodeTiwg
25

Is /clear worth it? Not for small, related tasks

The usual advice is to /clear between tasks so you stop paying for old context. We measured it, and for small related tasks the advice was wrong.

Article · @CodeTiwg
19

Paste the error, or describe it? Both found the bug; only pasted errors caught the second crash

When something breaks, you can tell Claude Code what you saw ("GET /notes returns a 500 for one user") or paste the actual error and stack trace. Pasting takes a few extra seconds of digging through l…

Article · @CodeTiwg
13

What a short CLAUDE.md actually changes: 3 runs with, 3 without

A CLAUDE.md earns its place when it tells Claude something it can't work out from the code. We tested that directly.

Article · @CodeTiwg
11
Load more