Thierry

SWE @ setup: omarchy, neovim, and pi (agent harness).

favorite model right now is gpt-5.5 and opus 4.8 for ui stuff

[vogel @ryanvogel]

woah this is sick

in july 5.6 Sol will be 750tok/s

[Adam @adamcohenhillel]

OpenPhone -

A new open source, customized Android OS

Built with AI from the ground up, and work on your existing old devices at home

[henrique cunha @henrycunh]

problem solvers, we're hiring at Tela.

we made a small experience for people who like messy, ambiguous problems.

it is intentionally under-specified. the data isn't perfect, the requirements aren't complete, and there's no clean scoring function.

that's real life right there.

a lot of the work here is figuring out what matters and understanding how Applied AI can be used to solve process-based knowledge work.

here you get the flow, the documents, a few stakeholders to talk to, and even a whiteboard.

you don't need a perfect answer. we care more about how you reason through it.

we want to see how you think and how you'd approach this use case. just click submit.

check it out at
fde dot tela dot com

the problems you'd get here are way more complex, and the solutions way more rewarding.

if building solutions for hard problems is your thing, we'd love to work with you.

[plannotator @plannotator]

Plannotator 0.21.2 is out

Custom agent reviews are here. Run @cursor_ai's famous thermo nuclear code quality and pstack interrogation reviews. Run your own reviews.

Code Review:
- Custom reviews link from your agent skills, run any skill as a review
- @cursor_ai + @opencode added as review providers

App:
- Codex Ask AI now works outside git repos
- Codex desktop now shows the review URL

[Thierry @0xthierry]

why did you choose to split the runtime and the sandbox? that seems to be the path most solutions are moving toward, but since you have your own built-in secrets proxy, couldn’t you run the model inside the sandbox, proxy the requests, and inject the secrets there?

am i missing something?

[Thierry @0xthierry]

i’ll share a few extensions i really liked and tweaked to work the way i like. some of these were inspired by other extensions, but i changed them to fit my workflow.

codex has the best compaction i’ve seen in the current harnesses, so i created this extension instead of using the default pi compaction prompt. this extension uses the openai compaction endpoint: https://t.co/bZEznZx4r2

i also created this amq-notify extension, which allows my pi to talk to / spawn claude code and have claude code messages injected automatically. it’s really good when i want to bring claude into the discussion, or when i’m running a goal and want claude to verify pi’s work: https://t.co/PLFBNqCyzR

the last one i think is worth mentioning is oracle, where i use my chatgpt subscription to access gpt-5.5 pro, so i can use a more powerful model when i need it: https://t.co/YpbKhwGEV6

then i have a bunch of other ones in there too...

[Emil Kowalski @emilkowalski]

Given how impressed I was with Cursor’s Thermo-Nuclear Code Quality Review skill, I decided to create a similar one, but specifically for animations.

/review-animations to review, and /emil-design-eng to build animations. They complement each other well

https://t.co/nbrXga5Btc

[dex @dexhorthy]

At HumanLayer, we’re on a mission to solve the AI slop code problem.

In 2025 we open-sourced our Research, Plan, Implement framework, now deployed inside fortune 500s like Block and Uber - places where shipping slop is just not an option

And that was just the beginning.

Today, we’re opening access to HumanLayer - an Agentic IDE, collaboration platform, and building blocks for your software factory.

HumanLayer enables engineers solving hard problems in complex codebases to:

> move 2-3x faster across the entire SDLC (not just coding)

> maintain rigorous standards for system architecture and program design

Hundreds of engineers at companies of all sizes are already using HumanLayer to ship fast without sacrificing quality.

I'm excited to invite you to try humanlayer today at https://t.co/cQ648EkrnG, and I'm even more excited to see what you build.

@0xblacklight and I are deeply grateful to our team, our customers who give us so much incredible energy and feedback, our investors who have always been in our corner, and our friends and family who have supported us along this crazy journey

if you're a staff or principal engineer trying to make AI coding work at scale for your team, we'd love to hear from you

as @swyx likes to say - let's make this the year of no more slop

[Unsloth AI @UnslothAI]

Gemma 4 12B can now run locally on just 8GB RAM via Dynamic GGUFs.

Google's new model, Gemma 4 12B Unified supports image, audio and 256K context.

You can run and train the model via Unsloth Studio.

GGUF: https://t.co/8cL321pVDh
Guide: https://t.co/odRo9WjRpA

[rafael thayto @thayto_dev]

#TIL global .gitignore

echo .DS_Store >> ~/.gitignore_global
git config --global core.excludesfile ~/.gitignore_global

[Tibo @thsottiaux]

Codex usage limits have now been reset across all paid plans. Enjoy the weekend!

[Lucas Meijer @lucasmeijer]

i never use skills. almost everything is a pi slash command instead.

[Thierry @0xthierry]

@davis7 @badlogicgames experiencing the same thing here

i had to add instructions to the appended system prompt to avoid this kind of behavior

[Thierry @0xthierry]

@thsottiaux @ajambrosino me either, specially the codex app on linux

[Mario Zechner @badlogicgames]

People of https://t.co/TgG5bkXUdV. If you've installed pi 0.73.1, you will now be notified to do a `pi update` for 0.74.0.

Starting today, all pi packages on NPM are in the earendil-works namespace, instead of mariozechner.

Update imports from @mariozechner to @earendil-works in extensions or SDK-based apps!

The pi GitHub repository has also moved to https://t.co/0h7Q5NmJCB

Update all the things!

[Thierry @0xthierry]

@thayto_dev Do you have auto compact enabled? Usually it consumes a lot of tokens

[Thierry @0xthierry]

please no 😿

[Armin Ronacher ⇌ @mitsuhiko]

People of pi. The issue tracker might take a vacation on the way to finish the great refactor, but we're still watching and fixing. Biggest changes are SSE downgrade for when websockets fail, rendering improvements to the read tool and some perf fixes for bash.

[henrique cunha @henrycunh]

github if OpenAI made it

[Thierry @0xthierry]

the pi /tree command is awesome, being able to go back and forth in a conversation is really useful, especially when fixing something