Jaka Hudoklin

@catalinmpit @samiralibabic Attack surface is mostly still your API tokens and credentials you give it access to and that's mostly unsolved problem...

@samiralibabic No. I want it as secure as possible since it has access to my personal stuff.

I've deployed my Hermes agent on a Hetzner VPS. The security measures I've taken so far: - Installed Tailscale and restricted SSH access to Tailscale IPs only - Blocked all ports except 80 and 443 - Restricted ports 80 and 443 to Cloudflare IP ranges only - Enabled and configured UFW - Disabled password SSH login Need to add: - fail2ban What else should I do?

Updated github.com/xtruder/openco… with Opus 4.8 support and few extra caching fixes. You can pretend you are ClaudeCode while using @opencode

@ggerganov not llama.cpp?

llama.cpp now has an official website: llama.app Our goal is to make local AI accessible to everyone, and improving the user experience is a big part of that. On the new landing page you’ll find a single-line cross-platform installer. The installation provides a single unified `llama` entrypoint which you can use to run/serve models and interface with 3rd-party agentic applications. While oriented towards simplified user experience, the new `llama` application also provides all the advanced functionality of the existing llama.cpp tooling with which experienced users are already familiar. Also note that all GGUF models that you might have already downloaded with llama.cpp in the past will be automatically available to use without downloading again (they are stored in the common HF cache on your machine). We have many improvements in the pipeline both at the UX and at the engine level and we plan to iteratively ship new things over the coming months. One of the main focuses will be seamless integration with local-friendly 3rd-party agents (such as Pi). In the meantime, we’ll continue to listen for feedback from the community and adjust accordingly, so keep letting us know what you think and need.

@TTrimoreau frame.work

Name a tech company that nobody hates

@AmayConnect @TTrimoreau I hope it's sarcasm 😂

@TTrimoreau Anthropic for sure!

@akramcodez Just wait a moment, need to accept call using terminal

the perfect smartphone for Linux obsessed nerds

@plainionist Why not both?

What brings you more joy? Coding with AI or coding without AI? 🤔

@YashHustle_22 5$ Opencode GO > 100$ Claude

Hot take: $200 claude > $20 codex > $100 claude

@KaiXCreator Boy it burnt tokens 🤑 🔥

wow. Anthropic just dropped Opus 4.8, and i asked it to refactor my entire codebase over 2 hours, 100 million tokens burned, complete architecture reset none of it worked but boy was it beautiful

@sflorimm I have maybe 5 agents open, but I was doing max like 3 running sessions

Does anyone actually run 10 agents in parallel?

@swyx New flag it seems: | `mid-conversation-system-2026-04-07` | Opus 4.8 only | Allows `role: "system"` messages mid-conversation |

"Developers can update Claude’s instructions mid-task without breaking the prompt cache or routing the update through a user turn" wtf? how??

@konshusfavorite @theo @fish_or_dog Opus 4.8 is money incinerator

@theo @fish_or_dog You keep lying about that. How much you getting from Scam Altman on this crusade of yours lately? Show the prompt, have people verify it independently. Let people prove what you are asserting. Lying and engagement farming while you getting paid for it is dishonest.

Resubbed at the $100 tier to try 4.8 and the new "ultracode" feature. I hit my limits in a single prompt.

@madsmadsdk Live a savage

I wrote code by hand today. Ask me anything.

@TinfoilTricorn @rohanpaul_ai OK Tinfoil

@rohanpaul_ai Heavy metals poisoning and birthdefects next.

🇨🇳 China’s doing a 2-layer setup: solar panels on the surface for electricity, and fish, shrimp, and crabs farmed underneath. Called Fishery-photovoltaic complementary" (FPC), integrates floating photovoltaic (FPV) solar panels with aquaculture.

While you hyping up Opus; Qwen 3.7 Max, Kimi 2.6 and Deepseek V4 are all great models that can solve your tasks for way cheaper. Stop hyping up Anthropic, they are overcharging you for inference and in return you are making them f*cking wealthy

@rezoundous Same cash burning machine, incinerator gives you batter cash burning value

If Opus 4.8 is not better than GPT-5.5, I'm gonna crash out.

Opus is like burning cash, especially when caching breaks, burnt through whole 20x 5h window in like 30min

@opencode Burning your money faster than ever before

Opus 4.8 now available in OpenCode

@serenaa_ge Nice work, can we also get a chart for average number of tokens required for each model to solve the tasks?

Today we’re releasing DeepSWE, a new standard for agentic coding benchmarks. On public leaderboards, top models often look relatively close in capability. DeepSWE shows where they actually diverge, reflecting the realistic experience of developers in their day-to-day work.

@TrustScore_1 That's a cope, we actually needed to solve problems ourselves where solutions were not on stack overflow or anywhere else.

NOBODY has written code from pure knowledge since 2008. Either developers Google'd it or use stack over flow AI just automated it.