Loading video...

Video Failed to Load

Go Home

Asking two agents in different harnesses to debug your code (from andirockk on IG)

611,385 views • 4 months ago •via X (Twitter)

32 Comments

Ansh's profile picture
Ansh4 months ago

Codex and Claude trying to fix bugs together 😂

Emmiti's profile picture
Emmiti4 months ago

@estrelloona and @jacobianmatrixx on my TL

Danny Shmueli's profile picture
Danny Shmueli4 months ago

So cute. Love the BSH on the left. Is the moral of the video that in the end, it works?

Justine Moore's profile picture
Justine Moore4 months ago

I hope so 🤞 I’m personally still in the trenches

Christina's profile picture
Christina4 months ago

It was a tough journey, but they have finally finished and delivered everything carefully.

iEarth's profile picture
iEarth4 months ago

Cats are cute no matter what they do, even when they're having a fight

Nirav's profile picture
Nirav4 months ago

within same harness also i experienced same thing. was running multiple claude code sessions to do review and fix on separate features but both agents kept committing each other's incomplete code, and interrupting each others tool calls somehow

Shalom Mann's profile picture
Shalom Mann4 months ago

I hate how much I like it

Chris Heatherly's profile picture
Chris Heatherly4 months ago

I feel seen

Michael Zellinger's profile picture
Michael Zellinger4 months ago

Has either of these cats raised their Series A yet? I’d like to slide in a small check

VIBECOBRA's profile picture
VIBECOBRA4 months ago

I don't use agents bc I had a curly bracket indentation error so I got VSCode copilot to fix. I think it was set to sonnet and it looped 8 times looking for error and still didn't fix the problem.

Eclipse 🌖's profile picture
Eclipse 🌖4 months ago

Two harnesses, two non-deterministic outputs — you’re basically paying for double the hallucination rate.

Renjit Philip 🔭💡's profile picture
Renjit Philip 🔭💡4 months ago

Good lord! This is exactly what happens when I ask my openclaw to check out why my Hermes agent failed

Arthur Wolsch 𝕏's profile picture
Arthur Wolsch 𝕏4 months ago

😮

Frankenstein (Hao)'s profile picture
Frankenstein (Hao)4 months ago

omg. the cats can cook.... 😂

Michelangelo's profile picture
Michelangelo4 months ago

A poor neighborhood went dry because you made that dumb video.

Roshan's profile picture
Roshan4 months ago

funny yar

Aalap Davjekar's profile picture
Aalap Davjekar4 months ago

Is this AI?

Aloysis Francis's profile picture
Aloysis Francis4 months ago

At end the code is broke 😂😂😁

CatGod's profile picture
CatGod4 months ago

Seeing two agents debug code side by side is WILD! What LENS are we looking through?!

JMoon's profile picture
JMoon4 months ago

cross-harness debugging works better than expected. Claude catches structural issues, the others get logic edge cases. different models see different failure modes

Dushyanth's profile picture
Dushyanth4 months ago

😂

Fabrizio Serafini's profile picture
Fabrizio Serafini4 months ago

@dhaber 👀👀

Amol Parikh's profile picture
Amol Parikh4 months ago

Debugging just became adversarial

Marco "Shikoba" Riccetti's profile picture
Marco "Shikoba" Riccetti4 months ago

True story 😂

Jim O'Shaughnessy's profile picture
Jim O'Shaughnessy4 months ago

🤣

King Louis XIII's profile picture
King Louis XIII4 months ago

@EdelaQuintaine & @JessfromW... 🤣🥰🤣

Veralll's profile picture
Veralll4 months ago

Haha, they messed up🤣

Utkarsh Singh's profile picture
Utkarsh Singh4 months ago

This is funny 😂

Roberts Cromwell's profile picture
Roberts Cromwell4 months ago

Keep sharing nice contents

Sanjeev Kumar's profile picture
Sanjeev Kumar4 months ago

Yay Justine got the Cat agents onto the job of grinding🤡😅

...'s profile picture
...4 months ago

@itsmeyayon payag ka love ganyan trabaho ni Woltir 😂

Related Videos

New short course: Building Code Agents with Hugging Face smolagents! Learn how to build code agents in this course, created in collaboration with Hugging Face, and taught by Thomas Wolf, its co-founder and CSO, and m_ric, Hugging Face’s Project Lead on Agents. Tool-calling agents use LLMs to generate multiple function calls sequentially to complete a complex sequence of tasks. They generate one function call, execute it, observe, reason, and decide what to do next. Code agents take a different approach. They consolidate all these calls into a single block of code, letting the LLM lay out an entire action plan at once, which can be executed efficiently to provide more reliable results. You’ll learn how to code agents using smolagents, a lightweight agentic framework from Hugging Face. Along the way, you’ll learn how to run LLM-generated code safely and develop an evaluation system to optimize your code agent for production. In detail, you’ll learn: - How agentic systems have evolved, gaining greater levels of agency over time—and why code agents are a next step. - How code agents write their actions in code. - When code agents outperform function-calling agents. - How to run code agents safely in your system using a constrained Python interpreter and sandboxing using E2B. - To trace, debug, and assess the code agent to optimize its behaviours for complex requests. - How to build a research multi-agent system that can find information online and organize it into an interactive report. By the end of this course, you’ll know how to build and run code agents using smolagents, and deploy them safely with a structured evaluation system in your projects. Please sign up here!

Andrew Ng

127,724 views • 1 year ago