
π©βπ» AI Content Developer β― @Andela β― ex @CommandCodeAI β― ex @hashnode β― ex @Rapid_API πΉ https://teapawt.net/t.co/U5nmjYnzdr
San Francisco // Vancouver
Joined August 2018
- Tweets1.3K
- Following249
- Followers962
- Likes2.5K
Claude Code or Codex is not for open source models
Seriously. Read this.
$1.
Thatβs all it took to make DeepSeek + Kimi K2.6 beat Claude Opus 4.7.
This isnβt a model breakthrough. Itβs a harness fix.
@CommandCodeAI is a coding agent harness for both open and closed models.
eng notes on our harness engineering below.
iβve stopped using Claude Code and Cursor
on top of that, the launch of Cursor 3.0 made that decision even easier.
Switched to Command Code
npm i -g command-code
with all the @AnthropicAI leaks
and after trying almost every coding agent out there
iβve come to one conclusion:
they all break the same way.
you hit limitsβ¦
so i've started using Command Code now!
a CLI based coding agent that learns your coding taste.
Itβs April 1st, and no, this isnβt a joke.
I didnβt think Iβd switch from Claude Code, but I did.
Iβm on @CommandCodeAI now.
It does everything other coding agents do, but the add-ons? Thatβs where it wins.
Terminal is about to feel πππππ in t-5.
Command Code > Claude Code > Cursor > Coding agents
1
5
212
Cursor, Copilot, & the other coding agents still kinda suck.
Use Command Code - coding agent with Taste by
@LangbaseInc
Launching soon π€
Cursor 2.0 is here β and a lot has changed!
Saw @leerob's post on it and I just had to try it out.
So I decided to test how good Cursor 2.0 really is by shipping a Support AI Agent using the Langbase SDK.