Michał Pierzchała

Principal Engineer for Open Source @callstackio. Core @reactnative Community contributor. Created agent-device

[Michał Pierzchała @thymikee]

@OrestesGaolin What for?

[Michał Pierzchała @thymikee]

HarmonyOS support for agent-device coming soon! we really mean it by being cross-platform
https://t.co/oZQJgJprLL

[Oskar @o_kwasniewski]

I will be talking about how we built a universal agentic testing platform!

see you there 👋

[Mike @grabbou]

Founder mode = activated. Setting up first round of discovery calls with companies that signed up for Apex beta.

If you missed it - do it now! ⬇️

LFG! 🔥

[Mike @grabbou]

Support for HarmonyOS ✅
First-time contributor ✅
Love Open Source

[Michał Pierzchała @thymikee]

initial support with only 3k lines is quite unexpected result (in a good way), we'll see where this will land.

cc @Huxpro this means supporting Lynx apps on HarmonyOS too

[Michał Pierzchała @thymikee]

ok I think the time has come for agent-device to mature from incubator to official Callstack GitHub org
https://t.co/L72ltVf2sY

[Lech Kalinowski @LeSiOO]

Give it one sentence + a deadline. It plans, scaffolds, installs native deps, runs the app on a simulator, screenshots itself, iterates on the visual bar, writes tests, and ships.

Brain: @Callstack's Apex LLM (it speaks React Native).
Hands: Nous Research's Hermes agent.

Unattended, in ~15–30 min each, it shipped a Skia semantic-graph note app with on-device TF-IDF linking, a shader glass-slab toy, and a breathing-orb guide — all verified on simulator.

When a build white-screened, it diagnosed the Reanimated/New-Arch bug, fixed it, and re-verified itself.

This is the future of mobile dev. 🧵👇

[Michał Pierzchała @thymikee]

@grabbou I'm waiting for these React Navigation tests to stabilize and I'm gonna do a killer post on recent reliability and perf improvements (all shipped with v0.16 btw)

[Michał Pierzchała @thymikee]

@RohitBa46914207 @LeSiOO @xeonbloomfield please take a look :)

[Michał Pierzchała @thymikee]

Say hi to 6 new models in our React Native Evals family:

#5 Claude Opus 4.8 → 84.7%
#11 Minimax M3 → 78.6%
#12 Kimi K2.6 → 77.7%
#14 Gemini 3.5 Flash → 75.5%
#17 MiMo V2.5 Pro → 73.4%
#23 Mistral Large 3 → 65.5%

Full benchmark ↓

[Callstack Engineers @callstackio]

We’re building an AI-assisted migration workflow to React Native.

It helps add React Native to an existing app, build the first proof of concept, and evaluate the ROI faster, with engineers still reviewing the work.

If you’re looking into migration, now is a good time to re-evaluate the first step.

Learn more from the article ⬇️ Link below

[Michał Pierzchała @thymikee]

@brandtnewlabs Very often just this. It’s only a verification tool after all

[Michał Pierzchała @thymikee]

boy I fixed so many bugs in agent-device thanks to Maestro being slow

[Michał Pierzchała @thymikee]

@gabimoncha @dmooralesg @opencode Gotta pick something

[Michał Pierzchała @thymikee]

@bharath__2020 the difference

[Michał Pierzchała @thymikee]

@RohitBa46914207 worth it?

[Michał Pierzchała @thymikee]

@Vapor_WwW @callstackio No, Composer 2 is Composer 2

[Michał Pierzchała @thymikee]

25 models evaluated so far, I guess I'll have to buy an ultra wide screen to fit them all

[Michał Pierzchała @thymikee]

https://t.co/t98iW8Z2qD

[Michał Pierzchała @thymikee]

@dmooralesg we have it internally but want to triple check as this is the only model that runs outside of @opencode harness.

hint: it performs better than Composer 2 in this benchmark