Dev Agrawal

He/Him
Content Creator
Engineer, @codemod
Core team, @solid_js
Core team, @ripple_ts

[Dev Agrawal @devagrawal09]

Everyone wants to do Code Mode agents

No one is truly embracing Code Mode

So I made a new agent with Code Mode as the foundation

Meet OC++ (a fork of @opencode)

https://t.co/ZJSS6vbSXR

[Dev Agrawal @devagrawal09]

The data loading and caching story of both ChatGPT and Claude are horrendous

I was really expecting better from trillion dollar companies but when you let AI drive the implementation without proper investment into infrastructure I guess this is doomed to happen

Please use a proper sync engine folks

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ Yeah I’ve spent a decent amount of time doing that the last few months

If you haven’t already I encourage you give solid 2.0 a try. You were probably using 1.x which might explain some of the issues. Agents are gonna have a much better time with 2.0

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ What kind of evals have you found helpful for Effect?

[Dev Agrawal @devagrawal09]

Like I said, I’d be more than happy to look over issues you’re running into and see how to fix them

The openclaw migration revealed a ton of issues that we then fixed and made Solid better for everyone

Feedback is more than welcome, it’s just that “agent confused solid for react” isn’t very helpful on its own

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ Yeah that’s my bad, here’s what I was trying to say in more detail https://t.co/ZYgQpbgbiI

[Dev Agrawal @devagrawal09]

That’s okay, my point has nothing to do with his work and more about the fact that a fleet of agents successfully performed a Solid migration without human oversight in a large existing codebase, which shouldn’t be possible if agents kept confusing solid for react as people think

I actually had my own agents go through the migration PRs and come up with a ton of improvements that make agents EVEN better at writing Solid

[Dev Agrawal @devagrawal09]

That’s okay, my point has nothing to do with his work and more about the fact that a fleet of agents successfully performed a Solid migration without human oversight in a large existing codebase, which shouldn’t be possible if agents kept confusing solid for react as people think

I actually had my own agents go through the migration PRs and come up with a ton of improvements that make agents EVEN better at writing Solid

[Dev Agrawal @devagrawal09]

@joelhooks @MichaelArnaldi @schickling @EffectTS_ How does that factor in here though? OpenAI doesn’t have a stake in Solid, in fact I’d imagine by this logic he would have used @octanejs whose author also works at OpenAI

[Dev Agrawal @devagrawal09]

Maybe instead of emphasizing the skill level of the hacker you should highlight how bad the security of the university infra should be to allow this

Didn’t even require Opus or Sol to break in

[International Cyber Digest @IntCyberDigest]

‼️ A low-skill hacker used a jailbroken DeepSeek V4 to breach an Indian public university and steal 5,200+ student and faculty records, Unit 42 says.

The AI agent even drafted a "NEXT EXPLOIT OPPORTUNITIES" plan.

Unrelated criminals now run the same cheap DeepSeek setup.

[Dev Agrawal @devagrawal09]

@joelhooks @MichaelArnaldi @schickling @EffectTS_ Why’s that

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ Definitely want feedback, please share!!

[Dev Agrawal @devagrawal09]

I understand not caring about the implementation details but that doesn’t make those details less important
The agent is dealing with the footguns instead of you, and every footgun increases token cost and likelihood of bugs
The reason you’d want to use Solid is the same reason you’d want to use Effect. More structural guarantees and less surface area for agents to mess up

[Dev Agrawal @devagrawal09]

@jpschroeder Except when they can’t because labs don’t want them to

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ If agents were bad at Solid this shouldn’t be possible https://t.co/MQ9bb3EvkY

[Peter Steinberger 🦞 @steipete]

Two weeks ago I kicked off a loop to move OpenClaw's Lit-based Web frontend to Solid 2.0.

Took ~6 days until we could remove Lit, and another week to move patterns to idiomatic Solid.

I checked in with the session every other day, overall the process was mostly automatic. Took about ~600 PRs.

Result: Smaller JS and CSS footprint and much more fine-grained update control. Everything feeling more snappy and responsive. https://t.co/rYAuaXH2xN

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ I’d love to learn more about these issues, if you can share the agent traces or maybe just ask the agents to look at those transcripts and write a small retrospective, it’d be massively helpful

[Dev Agrawal @devagrawal09]

@joelhooks @MichaelArnaldi @schickling @EffectTS_ As in you don’t care if the UI is good, or you don’t care how it’s built as long as it’s good?

[Dev Agrawal @devagrawal09]

@joelhooks @schickling @EffectTS_ Well we’re not done yet

[Dev Agrawal @devagrawal09]

@MichaelArnaldi @joelhooks @schickling @EffectTS_ I’ve been having agents write Solid for more than a year and I’ve yet to run into this problem

Maybe you have React specific skills that are polluting the context and confusing the agents?

[Dev Agrawal @devagrawal09]

@nkohari Kindof
There’s usually also a sync engine and a design system