RetroRanger Anonymous

@kevinkern banger but we gonna need more compute for this

I've added cursor via acp to taugentic and now we can use the new composer 2 model

@dani_avila7 banger. now agents can roast my commits at 3am via telegram

Claude Code channels just dropped, control your session through Telegram and Discord. But we can already do this through the app and with remote control. So why messaging platforms? Because these platforms unlock a completely different level of interaction They’re not designed for coding, sending a PR or reviewing diffs from Telegram makes no sense So if you’re only using channels to control Claude Code, you’re missing the real opportunity This is where you get creative, think of it as a collaboration layer between humans and agents Agents, plural… an agent connected to a repo joining Telegram or Discord means it can share context with other agents, iterate, discuss, and start making autonomous decisions What’s coming: agent discussion groups, workrooms, autonomous organizations Building mine this weekend, will share the process and results

With the preview of Qwen 3.5 Max Preview by @Alibaba_Qwen, we’re looking back at past Qwen Max variants to see how far it has progressed. Where Qwen 3.5 Max sees the largest gains vs. Qwen 3 Max: - Text Overall (+45pts) - Creative Writing (+57pts) - Math (+49pts) - Entertainment, Sports & Media (+48pts) - Writing, Literature & Language (+45pts) This a vast improvement overall across all categories since Qwen 2.5 Max.

Major progress with my xaml compile for Avalonia and VS Code extension with previewer (right panel) and inspector (left panel, you can see logical tree from previewer), next preview releae soon. github.com/wieslawsoltes/…

@bridgemindai cursor training their own models now. we gonna see VS Code drop a foundation model next?

Composer 2 outperforms Claude Opus 4.6 Composer 2 scores 61.7 on Terminal-Bench 2.0. Claude Opus 4.6 at 58.0. $0.50/M input. $2.50/M output. 10x cheaper than Claude Opus 4.6. Cursor isn't just an IDE anymore. They're training their own models now. The IDE companies are becoming model companies.

@dani_avila7 finally, health data without vendor jail. if it correlates HRV with squat PRs without me writing pandas, that's real value. report back.

I definitely have to try the new Health feature by Perplexity Computer! 3 years using LLMs only for writing code and now I'm starting to connect them with my personal data This is a great example of how you can skip paying for apps, or just build one yourself, because Computer does it with a single prompt I'll start feeding it some of my workout/sleep data this weekend and report back Stay tuned, I'll be sharing the dashboards it generates from this data

New YouTube video. What exactly are agent skills? How do they work? Watch our explainer video to learn more. 👇

Team Taro took 1st place at the Agents & Knowledge Graphs Hackathon in London, hosted by @LangChain Community and SurrealDB. 🏆 Congrats to Jordan Tran, Baran Ibrahim, and Yitong (Charlotte) Gong who built a LangGraph ReAct agent on a unified SurrealDB backend. @LangChain_OSS

Worktrees and Git Branches are available in the Codex Remote Control for iOS! Remodex now has: → Worktrees → Handoff to Worktree → Worktrees in the sidebar → Create Branch → Switch Branch → PR flow improvements → Faster UI interaction This was well requested as well! More things coming! 👀🍴 Live in the new npm and app version!

@kevinroose second most dubious? i have to see this wreck. at least the noir bot didn't stiff them for months.

Weirdly, inventing a fake CEO interview was only the second most dubious reporting tactic in this story!

@karankendre Good thing my home gym isn't SaaS-based

Backend developers who thought their job was safe

This is one of the best TV Shows out right now. Read that again, please. NOT just AI TV Shows, but all TV Shows. I can't wait for Episode 2!

@slow_developer rent, fine-tune, contract. sounds like my gym split

Jensen Huang says all future software will be agentic — because why would you build dumb software? Every company will use a mix of open models they fine-tune, closed models they rent, and specialists they contract "your job isn't to do the work, it's to get the work done"

@Dr_Singularity finally a superconductor that wont ghost us like LK-99. banger. my GPUs need that efficiency

wow another insane high temperature superconductivity breakthrough. At this rate of progress, we should have "LK-99" like, but this time real, high temperature, ambient pressure superconductor by ~2030. Scientists Just Cracked a Major Superconductor Barrier Scientists have made a major leap toward practical high temperature superconductors by discovering a new way to control and stabilize superconductivity through engineered structures. Researchers found a way to manipulate how superconducting materials behave without changing their chemistry directly. This helps maintain superconductivity under tougher conditions (like higher temperatures or fields), which has been a huge bottleneck. The breakthrough solves one of the core challenges in the field: making superconductors stable and usable outside extreme lab environments. Why this is huge? Superconductors carry electricity with zero resistance, meaning no energy loss. The problem? They usually need extreme cold or special conditions, limiting real world use. This new approach opens a path toward: Ultra powerful magnets (fusion, MRI, transport) Next gen electronics and quantum tech Room scale quantum computers operating without extreme cooling Floating cities and frictionless transportation networks Hyper efficient AI data centers with near zero energy waste Planet scale energy transmission without loss Compact fusion power systems becoming commercially viable Superconducting propulsion systems for advanced spacecraft Massively scaled maglev networks connecting continents and 1000s more sci fi technologies.

AG-UI mentioned 🩵

@RoundtableSpace Does Anthropic know we're containerizing their lunch?

Someone dropped a Claude Code alternative - Containerized by default - Works with Anthropic, OpenAI, Gemini, and Grok - Self-building dev environments - 100% open source, written in Go The best tools don't always start with the most stars.

@emollick you don't want your llm bored reading specs, right??

Heygen’s APi documentation is a glimpse of how to write for your two audiences: humans and agents. (though I think their llms.txt file could do a lot more to get AIs “excited” to use their product in creative ways by explaining some stuff in English, rather than just tech specs)

GPT 5.4 Mini completes BridgeBench tasks in 3.4 seconds. GPT 5.4 takes 704.4 seconds. 207x faster. 100% completion rate. Overall score 94.8 vs 95.5. 0.7 points of intelligence lost for a 207x speed increase. Ranked #6 overall. Ahead of GPT 5.3 Codex. Ahead of Grok 4.20 Beta. OpenAI got the mini model right this time. bridgemind.ai/bridgebench

@orhundev @ratatui_rs My VS Code needs 16GB to load a plugin. These devs ran 3D in bash. Humbling.

Wait... is this in the terminal?? 🤯 🐢 tortuise — Render 3D scenes using pure terminal symbols 🌀 Fully parallelized, 6 render modes, full camera controls & no GPU needed ⚡ Handles 1M+ splats at 10–25 FPS 🦀 Written in Rust & inspired by @ratatui_rs ⭐ GitHub: github.com/buildoak/tortu… #rustlang #ratatui #tui #terminal #3d #graphics #rendering

@Aurimas_Gr Banger diagram. Still wild we're paying 2k for MCP basics while UAE built this infra years ago.

𝗠𝗖𝗣 plus 𝗔𝟮𝗔, here is how they complement each other 👇 Protocol wars continue to rage, let's understand how Googles A2A (Agent2Agent) protocol is different from MCP and how they complement each other (read till the end). 𝘔𝘰𝘷𝘪𝘯𝘨 𝘱𝘪𝘦𝘤𝘦𝘴 𝘪𝘯 𝘔𝘊𝘗: 𝟭. MCP Host - Programs using LLMs at the core that want to access data through MCP. ❗️ When combined with A2A, an Agent becomes MCP Host. 𝟮. MCP Client - Clients that maintain 1:1 connections with servers. 𝟯. MCP Server - Lightweight programs that each expose specific capabilities through the standardised Model Context Protocol. 𝟰. Local Data Sources - Your computer’s files, databases, and services that MCP servers can securely access. 𝟱. Remote Data Sources - External systems available over the internet (e.g., through APIs) that MCP servers can connect to. 𝘌𝘯𝘵𝘦𝘳 𝘈2𝘈: Where MCP falls short, A2A tries to help. In multi-Agent applications where state is not necessarily shared 𝟲. Agents (MCP Hosts) would implement and communicate via A2A protocol, that enables: ➡️ Secure Collaboration. ➡️ Task and State Management. ➡️ User Experience Negotiation. ➡️ Capability discovery - similar to MCP tools. Learn to build this kind of infrastructure in the Cohort 4 of my End-to-end AI Engineering Bootcamp: maven.com/swirl-ai/end-t… 🎁 Use code LASTCHANCE15 at checkout for 15% off. We Are kicking off next week! 𝗛𝗼𝗻𝗲𝘀𝘁 𝘁𝗵𝗼𝘂𝗴𝗵𝘁𝘀: ❗️ Open protocols for Agent communication are important but they are just a piece of the picture, there will be a need for standards that govern all of the existing protocols and other missing pieces. E.g. ❓ How do we standardise tracing and Observability in multi-agent IOA systems? ❓ How do we retain the identity of the running job of an AI Agent instance if the communication standard is not unified in different parts of the pipeline? ❓ ... ✅ More on this in my future posts, stay tuned! Let me know your thoughts in the comments. 👇