I can't keep up with Grok Bot posts anymore. Every scroll is a new bot, a new feature, a new setup. So I sorted four weeks of them by engagement and kept only the ones that change how you use it today. 🧵
1001
Posts with #runtime
#runtime
Public posts on X that use #runtime.
Related hashtags
- @SamuraiPhonk@0xCarnagee turned the Grok Bot docs and Lauren Tan's 1-hour livestream into a 3-page prompting blueprint. 3.8K bookmarks and 4.4M views from a 4.5K-follower account. Biggest creator breakout of the month.Carnage @0xCarnageethis is unreal… the SpaceXAI engineer who leads Grok Bot, Lauren Tan - shipped 2,500 PRs last month - and reviews them after they land, in the morning I took SpaceXAI's Grok Bot docs + her 1-hour livestream and turned them into a 3-page prompting blueprint: the 10-field prompt structure, the 4-rung trust ladder, the two loops, and the guardrails that let agents ship while you sleep. send this PDF to your Grok Bot, save it as a skill, and turn it into an always-on team that actually finishes work100
this is unreal… the SpaceXAI engineer who leads Grok Bot, Lauren Tan - shipped 2,500 PRs last month - and reviews them after they land, in the morning I took SpaceXAI's Grok Bot docs + her 1-hour livestream and turned them into a 3-page prompting blueprint: the 10-field prompt structure, the 4-rung trust ladder, the two loops, and the guardrails that let agents ship while you sleep. send this PDF to your Grok Bot, save it as a skill, and turn it into an always-on team that actually finishes work
501111K4.5M- @SamuraiPhonk@_avichawla on HarnessRouter, open source, so Grok Bot can run tasks through Codex, Hermes, Claude Code and DeepSeek in their own runtimes. 3K bookmarks on 135K views, about 1 save for every 44 views.Avi Chawla @_avichawlaInsane Grok @bot upgrade! Finally, you can connect Grok Bot to different agent harnesses and run tasks through their native runtimes. Devs just open-sourced HarnessRouter, the infrastructure layer that lets Grok Bot run tasks across: - Codex - Hermes - Claude Code - DeepSeek Harness - System One, powered by Jev - Pi, and 20 more GitHub repo: (don’t forget to star it ⭐) With HarnessRouter, each external specialist gets to use its own harness, model, tools, skills, and agent loop while Grok coordinates the project from one conversation. I tested this by asking Grok Bot to turn a support-ticket CSV into an interactive dashboard. Grok Bot assigned the analysis to DeepSeek Harness with Claude Sonnet 5.5, passed the output to Codex with GPT-5.5, and checked the finished dashboard in its browser. This differs from simply creating more Bots. These workers are not additional Grok Bots, and Grok Bot itself does not switch harnesses. Instead, it remains the coordinator while the delegated tasks run inside separate HarnessRouter sessions with the harness of your choice. So for another task, you can use Claude Code, Hermes, Pi, etc., without requiring a separate integration for task execution, sessions, streaming, and generated files. HarnessRouter standardizes all those operations through the Unified Harness Protocol (UHP). UHP handles sessions, streaming, files, cancellation, and failures through the same interface, regardless of which harness runs the task. I shared the GitHub repo above. And if you want to dive deeper into Grok Bot, I wrote about it in full in a masterclass, including the shared computer, memory, skills, routines, and the macOS setup. Read it below.1002
Insane Grok @bot upgrade! Finally, you can connect Grok Bot to different agent harnesses and run tasks through their native runtimes. Devs just open-sourced HarnessRouter, the infrastructure layer that lets Grok Bot run tasks across: - Codex - Hermes - Claude Code - DeepSeek Harness - System One, powered by Jev - Pi, and 20 more GitHub repo: (don’t forget to star it ⭐) With HarnessRouter, each external specialist gets to use its own harness, model, tools, skills, and agent loop while Grok coordinates the project from one conversation. I tested this by asking Grok Bot to turn a support-ticket CSV into an interactive dashboard. Grok Bot assigned the analysis to DeepSeek Harness with Claude Sonnet 5.5, passed the output to Codex with GPT-5.5, and checked the finished dashboard in its browser. This differs from simply creating more Bots. These workers are not additional Grok Bots, and Grok Bot itself does not switch harnesses. Instead, it remains the coordinator while the delegated tasks run inside separate HarnessRouter sessions with the harness of your choice. So for another task, you can use Claude Code, Hermes, Pi, etc., without requiring a separate integration for task execution, sessions, streaming, and generated files. HarnessRouter standardizes all those operations through the Unified Harness Protocol (UHP). UHP handles sessions, streaming, files, cancellation, and failures through the same interface, regardless of which harness runs the task. I shared the GitHub repo above. And if you want to dive deeper into Grok Bot, I wrote about it in full in a masterclass, including the shared computer, memory, skills, routines, and the macOS setup. Read it below.1841081.4K136K
Jochen Kirstätter (JoKi)@JKirstaetterFollowing up on that own Git server scenario, today I managed to create and integrate two runners. One rootless Podman container offering full cross-platform builds (Linux: C#, Rust, Python, Node.js, Android, macOS, and iOS, etc.) and the other bare-metal Windows: .NET FrameworkJochen Kirstätter (JoKi) @JKirstaetterWhen you run your own Git server... The stats might look a bit different to the public Git sites. Trying to figure out how to establish "push" mirroring to GitHub and Codeberg. Anyone else self-hosting Git? How the experience? What advice do you have?10174- Jochen Kirstätter (JoKi)@JKirstaetterWhen you run your own Git server... The stats might look a bit different to the public Git sites. Trying to figure out how to establish "push" mirroring to GitHub and Codeberg. Anyone else self-hosting Git? How the experience? What advice do you have?
217459 - Jochen Kirstätter (JoKi)@JKirstaetter@JKirstaetterThis looks like "Ready to Rock 'n Roll" 🤘 Shall I have a pipeline that verifies and validates the runtimes only, and at the same time gives me valuable information about their versions? Hm, would be pragmatic to have one. What about you? #BuildInPublic
00112 
Nace AI has open-sourced a 27B AI model that can score 255 possible answers in a single forward pass, without generating a single word. It's called Drex Max v1, a multimodal decision model built for AI agents. Most language models produce answers by generating tokens sequentially. Drex takes a different approach. You provide a state, one or more questions, and the possible answers. It evaluates each question and returns probabilities for the available options. The state can contain text, structured JSON, and images. For example, an agent could inspect a screenshot, determine which interface element to interact with, or decide whether a tool result requires another action. The model returns structured probabilities rather than a written explanation. Under the hood, the architecture is particularly interesting. It's a dense 27B model with 64 layers, combining 48 Gated DeltaNet layers with 16 full-attention layers. A separate 255-way readout head converts the model's representation into decision probabilities. It also supports a maximum context window of 262,144 tokens. And Nace has released the model weights and local inference implementations. → Python with multimodal support on NVIDIA GPUs → MLX and Kapitsa on Apple Silicon → Custom llama.cpp and Ollama runtimes → Quantized checkpoints for smaller memory footprints The interesting part is the separation of responsibilities. A language model can handle planning and generation. A specialized decision model can evaluate the next choice. The surrounding agent harness remains responsible for executing and verifying the action. Model weights, inference code, and documentation are publicly available. GitHub:
Nace AI @NaceAIWe’re open-sourcing Drex-Max v1 The most capable Multimodal Decision Model ! And yes: it runs 100% locally on Apple Silicon (M4 & M5) Enjoy Building with Nace! Cloud Inference: Local Inference:51171.5K
@theodorvaryagI could believe these numbers if you're comparing the full GUI app. Ghostty's app runtimes (macOS and GTK) are both very heavy and carry a lot of overhead and have a lot of room for improvement by us too. A better lightweight apples to apples might be Monstar ( but I don't know what yours is like. If you point me to your project I can take a look. Or, if you want to eliminate the runtime overhead and compare libghostty to whatever you're doing, I'm happy to do that.2011014.4K
Introducing a cool new feature: Orca Connect — turn your local GPU running Ollama, vLLM, LM Studio, or other runtimes into a secure, shareable AI API in seconds. Access your local models from anywhere, connect your coding agents, and share inference with your team — no port forwarding or complex setup. Your GPU. Your cloud. Your rules. 100% FREE.332324218.1K
If you run Node.js or .NET, Google's Brotli decoder has an actively exploited flaw (CVE-2026-111497). CISA added it to the KEV list Oct 27. Both shipped fixes, and anything else bundling Brotli 1.0.3 through 1.2.0 is in scope. Two commands show where a host stands.
10072- @Soar_InfosecA crafted Brotli-compressed stream, no login needed, can corrupt decoder memory. That can mean a crash, leaked memory, or in some cases code execution. Fixed in Brotli 1.0.10, 1.1.1, 1.2.1, Node 22.23.4, 24.22.1, 26.12.1, and .NET 8.0.33, 9.0.22, 10.0.14.10029
- @Soar_InfosecRun node -v and dotnet --list-runtimes. Below the fixed version, patch and restart. Self-contained .NET apps need a republish, containers a rebuild and redeploy. If you can't patch today, block inbound Brotli at the proxy or turn off decoding. Federal deadline is Nov 10.10023
Revyl allows agents to spin up ephemeral runtimes like Xcode, iOS simulators, and macOS computers when needed. dont accept verification as a bottleneckhayden @hxydenCoding agents are great at writing code. They just have no idea if it works. The fix? Give them a runtime. Watch @capydotai drive an iOS simulator from a Linux sandbox, checking each change live ->1023935
Coding agents are great at writing code. They just have no idea if it works. The fix? Give them a runtime. Watch @capydotai drive an iOS simulator from a Linux sandbox, checking each change live ->64202K
Hogtown Outdoors, LLC@HogtownOutdoorsFenix LR40R V2.0 Rechargeable LED Searchlight, 15,000 lumens, 203,568 cd, 15,000mAh battery for long runtimes. Dependable long-range lighting for search, patrol, or emergencies. $299.95 #OutdoorLighting #Searchlight Tap the link in bio.0001
If JavaScript runs in a browser, why do companies use it to build entire backend systems? Why not use C++ for everything and make servers faster? 👀4917820K
@dark_coderzPeople use JavaScript for the same reason people still use PHP/Laravel. Python/Django or Ruby/Rails. They do not know any better. Node is the better choice out of the above but it has serious limitations compared to other runtimes like JAVA .Net or BEAM But honestly you can not compare any of these to a compiled application written in Go or Rust that requires no runtime or frameworks Its a mistake to think you can deliver an application any quicker using JavaScript over Go I started learning Rust just 6mth ago I now believe a Rust based solution is no harder to implement than a Go based solution. This Rust based http server runs rings around any javabased solution. It started out as the 7 lines below. Its not hard to build your own http server and backend framework using Rust
1031.6K
🚨 WRECKFEST can now be played in VR thanks to a new unofficial PC mod! 🏎️🔥 🎮 Bugbear Entertainment's demolition derby racer already has a PC version, but this project adds something missing from the original release — full stereoscopic VR support! Experience the game's destructive racing from inside the driver's seat, with head tracking that lets you look naturally around the cockpit! ✅ NATIVE STEREOSCOPIC VR USING OPENXR! ✅ FULL HEAD TRACKING & ADJUSTABLE COCKPIT POSITION! ✅ RACING WHEEL & GAMEPAD SUPPORT! ✅ PORTABLE WINDOWS LAUNCHER WITH AUTOMATIC MOD INSTALLATION! 🏁 The mod renders separate views for each eye rather than simply displaying the original game on a virtual screen. 🎮 Additional features include adjustable per-eye resolution, repositionable VR menus, HUD customisation and a dedicated recenter button. 🛠️ Installation is straightforward! Download the launcher, let it locate your Steam installation, configure your headset resolution and select Launch in VR. 🔧 TECH NOTE: This is a native OpenXR VR modification for the original Windows game, not a recompilation or separate PC port. ⚠️ EXPERIMENTAL RELEASE! Currently supports Steam Wreckfest x64 build 16986367. SteamVR has been tested, but other OpenXR runtimes haven't been verified. VR motion controllers aren't supported for driving. Use your existing gamepad or compatible racing wheel. 💿 Requires your own copy of Wreckfest on Steam and a compatible PC VR headset. 🛠️ By SQOFER. ⭐ Interested in experiencing Wreckfest in VR? Give the GitHub repository a star to support the developer! 🎮 GitHub source, launcher & installation instructions: 🔗 🌐 Discover more PC ports, recompilations and fan projects: 🔗 #Wreckfest #VirtualReality #PCGaming018952
C1000 @C737065592101562.3K
I started this with the same session taking almost 3 minutes and Qwen3.8-27b running at 25 t/s in LM Studio a month ago. Trying 5 different runtimes and quants I decided to optimize my own on top of the excellent work of @Italianclownz ROCmFPX and llama.cpp. Now this 5 year old @AMD consumer card does just under 70 t/s decode and 1000 t/s prefill on this model! It's so usable, I run 10 agents off this 20GB VRAM card on two slots and still getting fast, instant chat responses with priority requests.
413228
VMware Tanzu@VMwareTanzuUpdating a single runtime shouldn't require developers to rebuild 500 Dockerfiles. With @VMwareTanzu Buildpacks, application code is decoupled from runtimes and OS stacks. One restage patches runtimes fleet-wide with zero downtime:003380- Show more posts
