dunnne

[Ishita Zanje @dezinish13]

Postcards from Switzerland trip :)

[dunnne @undunable]

Bersiaplah loker buzzer rp. 500 per vote, vote kinerja presiden “Sangat Baik” di akun Threads dr. Gia

[dunnne @undunable]

With this simple screenshot, you'll know which model I'm currently using.

[dunnne @undunable]

Llm itu jawab prompt berdasarkan token prediction, jadi secara singkat ya llm sendiri gatau jawaban itu bener apa salah, tp dia jawab sesuai pattern saat dia latihan, ChatGPT sama Claude jg ngasih disclaimer kaya “Claude is AI and can make mistake”. weights/parameter llm jg sangat menentukan kualitas jawaban/output

[dunnne @undunable]

Medium Article: https://t.co/3E4SPVNSwU

@WalrusProtocol #MemoryWalrus

[dunnne @undunable]

Just submitted to the Walrus Hackathon Session 7.

Inkray Article:
https://t.co/b4UEkaXvwg

@WalrusProtocol
#WalrusMemory

[dunnne @undunable]

@WalrusProtocol Session 7 ends in 5 days;
there are 10 winners receiving $150 each, plus a $500 bug bounty program split among 5 people via GitHub issues (https://t.co/3T7YMG8Ju2).
The task is very simple: choose one of the six system prompts, polish it to suit your needs, and use it for a real-world project you are currently working on.

Use your polished system prompt in CLAUDE.md or AGENTS.md

Deepsurge: https://t.co/hdnlk85G5z

More infromation: https://t.co/3uEpuVAVlG

[dunnne @undunable]

Bench3xmaxxing

[Chris @ChrisGPT]

1. What.

[dunnne @undunable]

Omg

[Whales Premarket @whalespremarket]

🐳 NEW LISTING: $HEAT Pre-Market

Things are heating up. 🔥

@hibachi_xyz is building the home for stablecoin-native FX with a live orderbook, transparent pricing, and deep liquidity.

Trade your $HEAT allocation before TGE 👇

https://t.co/IQwnieXbdv

[dunnne @undunable]

a few months ago using Codex was like having unlimited tokens

[dunnne @undunable]

@hiarun02 I always ask ChatGPT to re-explain what Opus means

[dunnne @undunable]

@WalrusProtocol Session 7 ends in 5 days;
there are 10 winners receiving $150 each, plus a $500 bug bounty program split among 5 people via GitHub issues (https://t.co/3T7YMG8Ju2).
The task is very simple: choose one of the six system prompts, polish it to suit your needs, and use it for a real-world project you are currently working on.

Use your polished system prompt in CLAUDE.md or AGENTS.md

Deepsurge: https://t.co/hdnlk85G5z

More infromation: https://t.co/3uEpuVAVlG

[sandra @sandylikesfrogs]

finally we have achieved AGI

[Hugging Face @huggingface]

We've just surpassed 3 million models on the Hub 🤗

the community is accelerating towards an open, distributed future where open AI is everywhere, for everyone 🚀

[dunnne @undunable]

@rezoundous This is right, Opus 5 keep flagging unfinished work without even touch that work shit, in the end of the output ts always says “Remaining, most impactful first” i mean if that shit is impactful why dont work at it

[Tyler @rezoundous]

Opus 5 feels unlimited on Claude, but you can't do any serious work with it

[dunnne @undunable]

@mlbbfess panggil king

[Walrus 🦭/acc @WalrusProtocol]

As part of the Walrus Prompt Jam, we are spotlighting some of our favorite community projects built for portable agent memory.

First up: Markov by @/dun999 on Github.

Switching between Claude Code, Codex, and terminal agents usually means rebuilding prompt context from scratch. Markov preserves active task state across #AI coding tools so developers can hand off unfinished work instantly.

Who is currently building cross-agent dev tools or switching between AI coders? Drop your workflow below.

Explore the prompt: https://t.co/TLsNx9qiO8

View the full submission: https://t.co/YBQMtF3aQH

[Walrus 🦭/acc @WalrusProtocol]

Context stuffing is not long-term memory. RAG is not long-term memory.

Most agent frameworks rely on temporary workarounds that drop details, inflate token costs, or lock your state inside a single provider.

We mapped out how true agent memory is structured across semantic, episodic, and procedural state – and where today's tools break down.

Architectural breakdown here: https://t.co/l2BX2DHmdu

[dunnne @undunable]

@ZixuanLi_ Tibo but different