Meet PiG: @pidotdev, rebuilt in Go as one native binary.
Live multi-language extensions, derivative harnesses (Piglets), and everything Pi just in Go.
More at https://t.co/RtFt1dF1qQ
Announcing Gemini 4 Argon, our new frontier model.
Argon is built to sustain deep reasoning across complex, long-horizon workflows and delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.
We’re also expanding the model’s output token limit to an industry-leading 1M tokens.
Argon is currently rolling out to a set of trusted cyber defenders in the Fairwind Program, with broader availability as soon as possible.
Pi v0.99.2 is out!
Highlights:
- MCP servers with default codemode exposure no longer appear in the codemode description or block the first prompt; scripts use searchTools() and describeNamespace() to find them.
- New MCP auth options: oauth.clientName for servers that require known OAuth clients, and "auth": { "provider": "<provider>" } to authenticate HTTP servers with a provider's /login token.
- /reload now enables tools newly added to the defaultTools setting.
Complete details in thread ↓
building a coding agent? you can add herdr support yourself. no waiting on us.
new docs: https://t.co/wXOe9oMlEq
thanks to these folks for already shipping it 🫶
Chapel 2.10 is here! This release brings complete and productive union types, stack traces for thrown errors, and expanded public testing through GitHub Actions. Explore the highlights and see what’s new. #HPEDEV https://t.co/LjcRYfznwo
so the choices are:
- don't play by the rules, impersonate a sanctioned figma mcp client
- play by the rules, click on the link to the suggested form and find out that you can't play by the rules if you want figma mcp server access for your users
GPT-5.6 Sol → GPT-6 Sol → GPT-6.1 Sol (medium) on my mini-bench, in pi-agent and Codex
10 tasks from my own work × 3 runs, through Harbor.
The newer models solve every task here. My goal is to find the fastest, cheapest setup that still gets the job done without losing quality.
From GPT-6 → GPT-6.1:
> Codex: 20% fewer turns, 22% fewer tokens, and 15% lower cost.
> Pi used ~70% fewer tokens than Codex with GPT-6.1 (55k vs 185k per task), likely helped by its smaller system prompt and fewer turns.
What changed in trajectories? Batched Python checks instead of repeated CLI runs, whole-file reads, and fewer plan updates.
Side-by-side examples in the thread 👇
@dimhold Thanks Dmitriy! So Pi's behavior is the contract, bugs included. Parity tests diff every scenario against real Pi, and when Pi fixes a bug we move our pin to that release. We only diverge when matching is impossible/worse in Go. 🐷
‼️ PiG v0.3.0 is out ‼️
First big update since launch, thank you to the contributing herd and everyone trying it!
Now Pi extensions run Pi's own code in the Node runtime and much more in https://t.co/ggm8DcTlXz 🐷
🏠 https://t.co/RtFt1dF1qQ
Meet PiG: @pidotdev, rebuilt in Go as one native binary.
Live multi-language extensions, derivative harnesses (Piglets), and everything Pi just in Go.
More at https://t.co/RtFt1dF1qQ
@QiqiLabs 100%, pin both! If you're sharing a whole setup, a Piglet pins every package and builds it into one binary 🐷
pig install npm:[email protected] (or a git:…@ tag) and PiG won't load a package unless the installed version matches, same as Pi, and applies to all extension sdks
PiG v0.3.0 runs Pi extensions in Pi's Node runtime.
Before sharing a setup, I'd pin the extension versions too - not just the PiG release.
https://t.co/C22mkSZssn
🔈 if still running v0.2.0, re-run the installer once:
curl -fsSL https://t.co/xcy19TwqQ3 | sh
From 0.3.0 on, PiG tells you when a new version is out, like Pi does, and pig update handles it. Checksums and signatures are verified.
0.4.0 (Pi 0.99.1) coming soon. 🐷
‼️ PiG v0.3.0 is out ‼️
First big update since launch, thank you to the contributing herd and everyone trying it!
Now Pi extensions run Pi's own code in the Node runtime and much more in https://t.co/ggm8DcTlXz 🐷
🏠 https://t.co/RtFt1dF1qQ