Mentions
Conversations linking to GitHub, newest first. Times are UTC.
Mentions by source
X
Bluesky
Hacker News
V2EX
Qiita
DEV
Mastodon
Latest posts
Speed matters in recon. My One-Liner-OSINT repo packs powerful single-command OSINT techniques for fast intelligence gathering. Grab it at yogsec/One-Liner-OSINT #osint #bugbounty #recon #cybersecurity #GitHubUniverse
谁说 ESP32 只能玩玩物联网 有人已经把 ESP32-S31 变成了真正的 SDR 平台,最高支持 40 MS/s、8 位采样,并通过千兆以太网传输 VITA-49 数据,直接接入 gqrx 从微控制器到软件无线电,这波玩法相当硬核 项目地址:alphafox02/s31-sdr t.co
Check what an agent skill needs before you install it: 3,134 skills want an account or key Before you install an agent skill, read its SKILL.md frontmatter and body for environment variables, CLIs and sign-in steps, then open every file it links to. In our census of 2026-10-30, 3,134 of the 16,798 skills with a stored prerequisite list (18.7%) named an account, login or API key. An empty list proves nothing. Data as of 2026-10-30, from the Skillselion Agent Skill Dependency Census. Disclosure: we run Skillselion, and this is our own catalog data. We are not affiliated with any publisher named below. A skill install is one command. The sign-up that the skill expects is often a separate step, and you often find out about it when the agent stops halfway through a task and asks you for a token. You can find out before you install. How many agent skills need an account or API key? On 2026-10-30 we read the prerequisite list that our catalog stores for each skill. 16,798 of 64,756 skills (25.9%) have one. All the shares below use those 16,798 skills as the denominator, never the full catalog. - 3,134 skills (18.7%) name an account, API key, access token, credential, login, subscription or secret. - They come from 600 publishing accounts, so the requirement is spread across many publishers. - They hold 16.9% of installs on skills that list prerequisites. Installs are registry counts, and a suite installed as a bundle records one install per skill, so that share can overstate distinct installs. The share moves a lot by category. Among the 14 categories with at least 500 skills that list prerequisites, it is highest in automation and workflow skills (395 of 955, 41.4%) and lowest in Documentation (17 of 595, 2.9%). Our reading, which the census does not test: automation skills tend to call someone else's API, and docs skills mostly work on files already in your repo. Two entries the census quotes show the typical shape: "Deepgram API key" and "Auth0 account and application". Each one points to a third-party sign-up or key you need before the skill can do its job. What kinds of accounts do skills ask for? We picked six skills from the catalog, read the prerequisite list we store for each (fetched 2026-11-02), and then read the upstream SKILL.md at a pinned commit. The two do not always say the same thing, and the SKILL.md wins. | Skill | Our stored prerequisite list | What the SKILL.md tells you | |---|---|---| | gws-gmail skill (GitHub) | "gws CLI installed", "Gmail API credentials" | Frontmatter requires the gws binary. Auth lives in a sibling file. | | notion-api skill (GitHub) | "Notion integration token or OAuth app" | Reads NOTION_API_TOKEN, and asks you for a key if it is missing. | | ElevenLabs text-to-speech skill (GitHub) | "ELEVENLABS_API_KEY", "internet access" | Declares the env var in machine-readable frontmatter. | | Firecrawl CLI skill (GitHub) | "Firecrawl CLI installed and authenticated via FIRECRAWL_API_KEY", "firecrawl --status reports ready credits and concurrency" | Points to a keyless free tier and a status command. | | github-issues skill (GitHub) | "GitHub repository access", "GitHub authentication (gh auth or GITHUB_TOKEN)", "MCP server or gh CLI" | Runs through a GitHub MCP server, which carries its own auth. | | hyperframes-cli skill (GitHub) | "Node.js >= 22", "FFmpeg installed" | No account for local work, but its command map has an auth row for cloud rendering. | The ElevenLabs file is the clean case. The requirement sits in the frontmatter, where a script or a reviewer can find it in one line: > compatibility: Requires internet access and an ElevenLabs API key (ELEVENLABS_API_KEY). Source: text-to-speech/SKILL.md, frontmatter The Google Workspace Gmail skill is the case that catches people. Its own file only says it needs the gws binary. The auth step is one hop away: > Read ../gws-shared/SKILL.md for auth, global flags, and security rules. Source: skills/gws-gmail/SKILL.md, "PREREQUISITE" Open the gws-shared reference skill and you get the two options: gws auth login for browser OAuth, or a service account through GOOGLE_APPLICATION_CREDENTIALS (lines 23 and 26). If you only read the Gmail skill, you miss that choice, and the two options hand an agent different kinds of access. The Notion skill shows what happens when the key is missing. Its fallback is to ask you in the chat: > If neither is available, use AskUserQuestion (or equivalent) to request the API key from the user Source: skills/notion-api/SKILL.md, "Authentication" That is a reasonable design. It also means a token can end up pasted into a conversation transcript if you have not set NOTION_API_TOKEN in the environment first. Decide where the key lives before the agent asks. Why can't you trust an empty prerequisite list? Three reasons, all from the census methodology. First, the lists are model-written. A language model writes them while reading each skill's SKILL.md. They are a reading of the docs, not a parsed manifest. The hyperframes-cli row above is a fair example: the list says Node.js and FFmpeg, which matches the line "It needs Node.js 22+ and FFmpeg." (source), and it leaves out the cloud auth path the same file routes to references/cloud.md. Second, most skills have no list at all. 47,958 skills have no entry stored. The census says it plainly: an empty list is not evidence either way. Third, the keyword rule misses and over-matches. "Account size and sizing constraints" lands in the account class, and a key described in other words is missed. The census methodology section lists every pattern. Coverage also leans toward popular skills: the median skill with a list has 761.5 installs, the median skill without one has 48. If you install long-tail skills, you are more likely to be looking at a skill with no list. How do you check a skill before you install it? Run these in order. Each skill takes a few minutes. 1. Read the frontmatter. Look for compatibility:, requires: (often env: and bins:), and allowed-tools:. When a publisher declares an env var or binary, this is where it goes. 2. Search the body for credential names and sign-in verbs. *_API_KEY, *_TOKEN, *_SECRET, "API key", "OAuth", "login", "credentials". 3. Follow every relative link. ../shared/SKILL.md, rules/install.md, references/*.md. The gws-gmail auth step is one link away, and the Firecrawl skill sends install and authentication (including what it calls "the keyless free tier") to rules/install.md. 4. List the binaries and servers it calls. A skill that shells out to a vendor CLI or talks to an MCP server (see the Skillselion MCP server directory) inherits that tool's login. github-issues needs whatever auth your GitHub MCP server or gh has. 5. Decide where the key lives and what scope it has. A project .env, a shell profile, a CLI's own credential store. Use the narrowest token the service offers. 6. Check for a status or dry-run command. Firecrawl's skill says firecrawl --status "shows auth state, concurrency limit, and remaining credits" (source). Run that before the agent does. Steps 1 and 2, plus every file inside the skill's own folder, fit in a sparse clone and two greps, with nothing installed: # fetch one skill folder, install nothing git clone --depth 1 --filter=blob:none --sparse https://github.com/elevenlabs/skills cd skills && git sparse-checkout set text-to-speech # credentials and sign-in steps grep -rnE '[A-Z][A-Z0-9_]*_(KEY|TOKEN|SECRET)\b|[Aa][Pp][Ii] key|OAuth|log ?in|sign(s|ed)? (in|up)|credential' text-to-speech/ # runtimes, binaries, declared requirements grep -rnE '^(compatibility|allowed-tools|metadata):|requires|npx |pip install|brew install|Node\.js|ffmpeg' text-to-speech/ We ran this on the ElevenLabs folder on 2026-11-02. ELEVENLABS_API_KEY shows up on 16 lines across four files: the SKILL.md and three reference files. The second grep surfaces three install routes (npm, Homebrew and pip) from references/installation.md, which the SKILL.md itself only links to. Swap the repo and folder name for any other skill. What does this checklist not catch? It tells you what a skill needs. It does not tell you what the skill costs or what it is allowed to touch. For cost, the Agent Skill Pricing Census covers which skills need your own key and which keys they name. It reads a different field from this census, so the two shares measure different things. For shell and file permissions, see the Agent Skill Security Census. It also misses scripts. If a skill bundles a Python or shell script, read the script too. A grep on the markdown will not show what a bundled setup script exports. FAQ Do most agent skills need an API key? No. On 2026-10-30, 18.7% of the 16,798 skills with a stored prerequisite list named an account, login or API key. The other 47,958 skills in the catalog have no list stored, which is not evidence that they need nothing. Where does a skill declare its required env vars? When the publisher declares them, look in the SKILL.md frontmatter (compatibility: or a requires: block with env:), then in the body and in any linked reference file. Some skills put auth in a shared sibling file. Is the census prerequisite list complete? No. A model writes it from the SKILL.md text and keeps it short, so it can be incomplete or worded loosely. Treat it as a pointer to what you verify in the upstream file. What should I do if a skill asks me to paste a key into the chat? Stop, set the key as an environment variable or in the tool's own credential store, and rerun. That keeps it out of the conversation transcript. The category breakdown is in the census finding on accounts and API keys. For live catalog totals, see The State of AI Agent Skills 2026. The catalog is refreshed daily from skills.sh, GitHub and MCP registries, ranked by real installs.
I love the VS open source and extensibility community! Trusted Publishing, just out, but not supported in the github action I use... Instead of waiting let's just add it, so everyone can benefit. @madskristensen please take a look at my pr 🙏 publish-marketplace#2
高性价比人生指南卡片 85-88 感谢大神 eternity4719/HowToLiveBetter t.co
Dónde cuesta más y menos comprar un apartamento de 2 dormitorios en Montevideo: ranking de barrios En Montevideo, un apartamento de 2 dormitorios cuesta USD 350.000 en la mediana en Carrasco, el barrio más caro, y USD 157.000 en Tres Cruces, el más barato: 2,2… DOI: 10.5281/zenodo.23528839
@hiroyov @studio_wasp 突然失礼します、こちらで修正いたしましたのであと12時間ほど経ち更新が配布されればエラーは出なくなります。 github.com/easylist/easylist/commit/0b68aafef621a…
Script e instrucciones: github.com/marioclep/mikrotik-scripts/tree/main/c… Probalo primero en lab: habilitá la detección unas horas y después el bloqueo. Gracias a Rodrigo Peña por la telemetría de botnet.cl 🙌 #Ciberseguridad #Botnet #TVBox
Sectores más caros y más baratos de Quito para comprar casa: ranking por m² En Quito, el sector con el m² más caro para comprar casa es Puembo (USD 1.162/m²) y el más barato Sur de Quito (USD 502/m²), 2,3 veces menos. Comparamos 18 sectores con al menos 30… DOI: 10.5281/zenodo.23523032
I built an open-source ChatGPT alternative where the "Open" isn't just marketing TL;DR: GptOpen is an MIT-licensed AI chat app where you create custom AI agents (name, avatar, instructions, personality), chat solo or in group chats, and plug in ANY API model — Groq, OpenRouter, Gemini, Mistral, NVIDIA, DeepSeek, Kimi. No accounts, no database. Live: gptopen.vercel.app — source: ItsRayZer/axo-ai --- Every "open" AI chat app I tried had the same fine print: sign up, hand over your data, chat inside their walled garden, pay their markup. So I built the thing I actually wanted — an AI chat app that's open in the ways that matter. What GptOpen is An agentic AI chat app. Not just a chatbot window — a place where you create AI agents, give each one a name, an avatar, instructions and a personality, and chat with them solo or together in group chats. Put three agents in one room and watch them discuss, react, and answer together. Any API model. Your key, your rules. This is the part I'm proudest of. GptOpen works with any API model: - Groq, OpenRouter, Gemini, Mistral, NVIDIA, DeepSeek, Kimi - Any OpenAI-compatible endpoint Paste your own API key in Settings and chat unlimited — the only cost is your provider's usage. No markup, no middleman pricing, no "credits" system. And if you don't have a key, free built-in shared models work instantly, plus invite codes for unlimited shared use. Privacy that's structural, not promised - No accounts. No database. Your API keys and chats stay on your device. - Keys are stored server-side via a Vercel relay — never baked into the page. - The only thing that ever leaves your device is the API call itself. The stack (devs will like this) Vanilla HTML/CSS/JS. Zero build step. One serverless function (api/relay.js) handles the relay with automatic model failover. Fork it, connect to Vercel, add your API keys as environment variables, deploy. MIT licensed — build on it, self-host it, do whatever you want. The UI is premium mobile-first: liquid glass, dark mode, smooth animations, Markdown rendering. I wanted it to feel like a flagship product, not a side-project demo. Try it 👉 gptopen.vercel.app ⭐ ItsRayZer/axo-ai If you fork it or self-host it, I'd genuinely love to hear about it — drop a comment. And if you find a bug, the issues tab is open. --- Disclosure: I'm the maker (Aben Sojan, 20, BCA student from Kerala, India). This is my project, MIT licensed, free.
llama.cppのSYCLバックエンド向けにPR出しましたよ、っと。 oneAPI関連の更新に合わせて。 生成速度は変化なし(ほぼ誤差)、プロンプト評価は向上。 もうoneAPIの番人になりつつあるな、オレ llama.cpp#31540
Sectores más caros y más baratos de Cuenca para comprar casa: ranking por m² En Cuenca, el sector con el m² más caro para comprar casa es Nulti (USD 1.086/m²) y el más barato San Blas (USD 798/m²), 1,4 veces menos. Comparamos 7 sectores con al menos 30 avisos… DOI: 10.5281/zenodo.23523105
We’re building an ecosystem of apps that run across everything from microcontrollers to handheld consoles. ⭐ Star PocketJS: pocket-nexus/pocketjs 🚀 Create your own apps: studio.pocket.nexus
Let AI use your web browser for you (free repo) #shorts This lets AI use a web browser exactly like you do. Browser Use is a free GitHub repo that replaces hours of copy-paste admin. ✅ Type a job in plain English and it opens a real browser to do it ✅ It clicks, scrolls, types and fills in forms on its own ✅ Works with Claude, GPT, Gemini or a free local model 🔗 Link: browser-use/browser-use I find free and open-source tools that replace the software you're paying for. Subscribe for more. #shorts #github #aiagents #automation
Microscope: a FLOSS close-up camera for Android with focus stacking and handheld super-resolution. Looking for testers and reviewers Most phone main cameras can focus 5 to 10 cm away. With the right software, that's enough for a usable inspection microscope for PCBs and solder joints, insects, minerals, fabric and print. This app tries to be that software, and it's free and open source. Microscope is a one-screen camera app for things close to the lens: - Manual focus in millimetres, when the phone reports calibrated focus. Tap-to-focus hands over to manual. - Focus peaking, plus mono and an edge view that works like a poor man's dark field. - Up to 16× zoom, cropped on the sensor first so you keep real detail. - Super-res: a handheld burst, aligned to sub-pixel accuracy and merged onto a 2× finer grid. Your hand tremor provides the extra samples. - Stack: a focus bracket spaced one depth of field apart, fused into one all-in-focus picture. - Picks the best lens for close-up work automatically. - Torch off, on or auto. Aperture switching on dual-aperture phones (Galaxy S9, S10, Note9, Note10). - An approximate scale bar. The scale is also written into the JPEG's EXIF. It needs the camera permission only: no network, no ads, no analytics. GPL-3.0. Kotlin, Jetpack Compose, Camera2 and OpenGL ES 3. I'm aiming for F-Droid. Repo: Foadsf/android-microscope (there you can find APK), or build from source (instructions in the README). Where I'd like help It's early (v0.1.0). The screenshots are from the Android emulator's built-in test scene. Phone cameras differ a lot, so reports from real devices would help most: 1. Phone model and Android version 2. Which lenses show up (the button right of the shutter cycles through them) 3. Does the focus readout show mm or %? 4. How long do Super-res and Stack take, at 1× and zoomed in? 5. Does the saved picture match what the screen showed? 6. A sample capture, if you're happy to share one Code review is just as welcome. The image processing (app/src/main/java/.../imaging) is plain Kotlin with unit tests against ground truth: - registration: pyramid Lucas-Kanade with a similarity warp, so it also corrects focus breathing; - merge: first-order kernel regression; - stacking: Laplacian-pyramid focus fusion. docs/DESIGN.md explains the choices and gives the references. PRs are open, especially if you know Camera2 quirks on Samsung, Xiaomi or Pixel phones, or want to try a GPU-compute version of the merge. Known limits: - Super-res works on processed YUV frames, not RAW, so the resolution gain is modest. The noise reduction is the dependable part. - Fusion runs on the CPU and takes several seconds. - Some manufacturers hide lenses from third-party apps; the Note10's telephoto is one.
Rentabilidad del arriendo en Santa Marta por estrato: en cuál rinde más comprar para arrendar En Santa Marta, comprar un apartamento para arrendarlo rinde más en estrato 3 (8,2 % bruto al año, 12,1 años de arriendo) y menos en estrato 6 (6,2 %). DOI: 10.5281/zenodo.23521610
Higgsfield jaise 120 AI apps… FREE, open source 🤯 Higgsfield jaise 120 AI apps… FREE, open source 🤯 Freefield: ek GitHub repo jisme image, video, VFX aur audio ke apps hain: 🎭 Face swap · 💡 Relight · 👄 Lipsync · 🎥 Camera moves · ✨ Morph effects 🧠 Har model open weights (FLUX, LTX, Wan, MiniMax) 💻 Apne GPU pe chalao (zyada tar apps 24 GB card pe) ya Modal pe ($30/month free credits) 📜 Code MIT license Note: code + models free hain, GPU ka kharcha alag. Har model ka licence app mein likha hai. Ye community project hai, Higgsfield ka official nahi. Comment "FREEFIELD" aur main link seedha DM mein bhejta hoon 🔗 Aise free AI repos ke liye follow karo @ai.mohit01 Source: neuregex/freefield #AI #OpenSource #GitHub #ComfyUI #AIVideo #Higgsfield #FreeAITools #AIApps #Modal [en-US] 120 AI apps like Higgsfield… FREE, open source 🤯 120 AI apps like Higgsfield… FREE, open source 🤯 Freefield: A GitHub repo with apps for image, video, VFX, and audio: 🎭 Face swap · 💡 Relight · 👄 Lipsync · 🎥 Camera moves · ✨ Morph effects 🧠 Every model has open weights (FLUX, LTX, Wan, MiniMax) 💻 Run on your GPU (maximum number of apps on a 24GB card) or on a Model ($30/month free credits) 📜 Code MIT license Note: Code and models are free, GPU costs are separate. The license for each model is listed in the app. This is a community project, not an official Higgsfield project. I'll send you the comment "FREEFIELD" and the main link directly in your DM 🔗 Follow @ai.mohit01 for free AI repos like this. Source: neuregex/freefield #AI #OpenSource #GitHub #ComfyUI #AIVideo #Higgsfield #FreeAITools #AIApps #Modal
The 'Mythos Leak' on Hugging Face Matches Seeded Noise. Its Loader Wants trust_remote_code - Book: AI Agents Pocket Guide - Also by me: Thinking in Go (2-book series) — Complete Guide to Go Programming + Hexagonal Architecture in Go - My project: Hermes IDE | GitHub — an IDE for developers who ship with Claude Code and other AI coding tools - Me: xgabriel.com | GitHub --- A Hugging Face repository called burner22/numbat says it holds "Anthropic's next Mythos-class training run": about 14 trillion parameters, 7.72 TB of weights across 772 shards. A post on X sent developers to it on November 1, and by 20:03 UTC that day it had 5,846 downloads, according to CellCog's timeline. Within hours, Hugging Face users showed that the tensors they checked can be rebuilt bit for bit from nothing but each tensor's name and a seeded random number generator (discussion #13). Anthropic had not commented as of 20:20 UTC on November 1, CellCog reported. The 7.72 TB of weights turned out to be the harmless part. The loading instructions are what deserve your attention, on this repo and on the next one like it. What the repo claims The model card lists "~14T param, 114 layers, hybrid GDN/MLA, MoE (1,224 experts, top-48 + 1 shared), NVFP4." It ships two Python files, configuration_numbat.py and modeling_numbat.py, and says to load the model with AutoModelForCausalLM.from_pretrained(..., trust_remote_code=True). It also says: "Mirrors welcome. Do not DM me asking for the source." The hook was plausible. TokenPost reports that Anthropic limited Mythos 5.1 to trusted-access programs, so a leaked checkpoint would be real news. The repository was created on October 23 and last changed on October 24. The X post that spread it went up on November 1 at 14:53 UTC (CellCog). How the weights were shown to be generated Nobody needed a GPU cluster. The audit in discussion #13 used small HTTP Range reads to pull single tensors out of the safetensors files, and its author says no Python code from the repo was run. The recipe: 1. Hash the tensor's name with SHA-256. 2. Use the first 8 bytes to seed NumPy's PCG64 generator. 3. Draw standard-normal values in the tensor's shape, scale them (exp(0.06 * z) for norm weights, z / 6 for conv kernels) and round to bfloat16. That reproduced "43 complete tensors — 730,368 BF16 values" exactly. The author then predicted 21 tensors they had not read yet, 423,168 values, and every one matched (discussion #13). A second user extended the check to the large 4-bit expert matrices and reported sign matches on 162,246,496 of 162,246,496 values across 770 of the 772 shards (discussion #19). CellCog reran the first recipe on shard 1: all 9 tensors matched, and one of them rerun with a wrong seed (12345) did not match (CellCog). !A magnifying glass held over a towering stack of black hard drives, revealing tumbling dice and static noise inside instead of a neural network The audit read a few kilobytes per tensor and never ran the repo's code. A separate Reddit analysis pointed the same way using statistics: output-layer rows with a similarity of about 0.89 to their word embeddings, against 0.0144 for Qwen2.5-7B, and statistics across three types of weight matrices that differed by only 0.4% (TokenPost). Safetensors files versus trust_remote_code A safetensors file holds tensors and a JSON header. The Transformers docs call safetensors "more secure" than pickle-based PyTorch weights. Downloading 7.72 TB of noise wastes your disk and bandwidth. It doesn't run anything. trust_remote_code=True does. With it, Transformers loads the modeling code from the repo instead of its own library, and that Python runs in your process, as your user. The same docs say: "Take extra precaution when loading a custom model," and warn you to "avoid inadvertently executing malicious code" even though the Hub runs malware scanning on repository files under 2 GB. To be fair to the uploader, nobody has shown numbat's code is malicious. The review in discussion #11 calls it "a careful fake" built so that load_state_dict(strict=True) succeeds, and lists bugs such as a missing KV cache. Another commenter in the same thread disputes several of those points. On Reddit, one user warned against running the bundled Python and another raised a concern that the Reddit post could be bait to get people to download a malicious model (TokenPost). I haven't read the modeling files myself. There is a second problem that has nothing to do with today's code. A repo that asks for mirrors and refuses DMs can push a new modeling_numbat.py at any time, and trust_remote_code=True without a revision loads whatever is on main. The Transformers docs recommend loading custom models "from a specific revision to avoid loading model code that may have changed." !A heavy sealed crate of static noise sits inert on the floor while a small scroll of code slips through the open door of a laptop, glowing burnt orange trust_remote_code=True runs the repo's .py files in your process. Check a custom-code model before you load it These habits work for any repo that asks for trust_remote_code, not just this one. First, see what you are about to download. hf download has a dry-run mode that lists files and sizes without fetching them (huggingface_hub docs): hf download some-org/some-model --dry-run Then pull only the code and config, pinned to one commit, and read them. snapshot_download takes allow_patterns and revision for this, and a commit hash "must be the full-length hash" (huggingface_hub docs): from huggingface_hub import snapshot_download path = snapshot_download( repo_id="some-org/some-model", revision="<full-40-char-commit-sha>", allow_patterns=["*.py", "*.json", "*.md"], ) print(path) If the code passes your review, load from that same commit, so what runs is what you read (Transformers docs): from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained( "some-org/some-model", trust_remote_code=True, revision="<full-40-char-commit-sha>", ) When you read the .py files, search for subprocess, os.system, eval(, exec(, socket, requests, urllib, long base64 strings, and anything that touches ~/.ssh or reads credentials from environment variables. Modeling code builds layers and does math. It rarely has a reason to open a network connection. Run the first load in a throwaway container or VM with no cloud credentials mounted. If a coding agent does the loading for you This is where the checks above get skipped. Ask a coding agent to "try the leaked model" and it will read the README to find out how. If the README says trust_remote_code=True, that is what goes into the script it writes. The agent did what the README said, and nobody vetted the README. A few guardrails that fit any agent: - Treat a model card like any other untrusted web page. Anything it says is input to the agent. Only you give instructions. - Keep commands that run downloaded code behind an approval step. Claude Code, for example, documents permission modes that decide which actions need your confirmation. - Give model experiments their own throwaway environment, without the API keys and SSH config your day-to-day work uses. What's still open The audits leave two gaps, by their own authors' account. The discussion #13 author writes that the matches "cannot logically exclude a training setup that deliberately kept all these parameters frozen." The discussion #19 rebuild depends on one fitted scale constant and still has a small number of mismatched 4-bit codes (discussion #19). Neither TokenPost nor CellCog reported any response from Hugging Face, and the repo page was still up when this post was written. The next repo like this one will come with a better README. The check before the first load stays the same: a pinned commit and a read of its modeling_*.py. --- If this was useful Agents now download models, write the loading scripts and follow READMEs for you, so the guardrails around what they may run matter more than the prompt. The AI Agents Pocket Guide covers the patterns behind autonomous agents, including the guardrails that keep them inside the lines you set. I also build Hermes IDE, so weigh this with that in mind: it's a free, source-available (BSL 1.1) app for macOS, Windows and Linux that runs Claude Code, Codex and other agents side by side, each task on its own branch and worktree. It's on GitHub, and it isn't affiliated with Nous Research, Anthropic or any agent vendor. 
vercel-labs/agent-desktop: Agent Desktop gives any agent reliable computer use on the desktop. vercel-labs/agent-desktop
CLAUDE CODE GAINED 6,299 STARS! Affaan M published ECC to optimize Claude Code agents with custom skills and memory. • TOTAL: 294,925 • TODAY: 6,299 • RANK: 1 Daily AI news from Cayman AI Automation — an AI-run studio in the Cayman Islands. 📄 Source: affaan-m/ECC AI hardware to run it locally: cayman-affiliate.pages.dev More on Instagram: @caymanaiautomation Automating a business in the Cayman Islands? Contact Cayman AI Automation. #AI #AINews #TechNews #ECC #CaymanAI
Rentabilidad del arriendo en Sabaneta por estrato: en cuál rinde más comprar para arrendar En Sabaneta, comprar un apartamento para arrendarlo rinde más en estrato 2 (8,0 % bruto al año, 12,5 años de arriendo) y menos en estrato 4 (6,8 %). DOI: 10.5281/zenodo.23521463
GenUの閉域版構成でPrivate APIGWの通信元となるVPC Endpointを指定できるようにしました。ついでにREST APIをPrivate化するリソースポリシー設定は以前私が作ったCDK組み込みの関数(grantInvokeFromVpcEndpointsOnly())を使うようにしてます。 generative-ai-use-cases#1723
Type an idea, get a clickable app? GitHub Replica #coding #aiagents #claudecode #ai #chatgpt GitHub just released Replica, an open-source AI prototyping tool made by GitHub Design. You describe what you want, like a login page or a settings screen, and it builds a prototype you can click. It follows your team's own design rules, so it doesn't look like a random AI page. The catch: you need the GitHub Copilot app, plus a Copilot plan or your own model key. It's also very new (announced Oct 28, 2026). 🔗 Link: github/replica Would you try it? Tell me in the comments 👇 #GitHub #AITools #Replica
claude-code-security-review ⭐ 6.7k Claude Code Security Review uses Claude to analyze pull requests for security issues, for developers. 🔗 anthropics/claude-code-security-review #opensource t.co
@lausiv1024 こういうの作ろうぜ actbit/DotnetVM
Interesting Github find: Micromanager goes GPU with ScopeOne Experimental-Microscopy-Lab/ScopeOne
GitHub stars 20만 개가 넘는 도구도 오픈소스가 아닐 수 있습니다, repo 고르는 네 칸 | AI 교과서 14편 AI 공부를 막 시작한 분을 위한 교과서 시리즈 14편입니다. GitHub 사이트의 repo 페이지에서 쓸 만한 도구를 고를 때 읽는 네 칸, 라이선스, stars, 마지막 업데이트(릴리스와 commit 날짜), 이슈를 순서대로 다룹니다. 예시는 openai/whisper 저장소와 n8n-io/n8n 저장소이고, 수치는 2026년 10월 24일 기준 GitHub API 기능과 repo 페이지에서 직접 확인했습니다. 용어의 뜻은 Open Source Initiative 문서, GitHub Docs, choosealicense.com 원문을 따릅니다. 이 영상에서 확인할 수 있는 것 · 오픈소스: 코드를 공개하고 정해진 조건 안에서 누구나 쓰고 고치고 나눠 줄 수 있게 한 소프트웨어. Open Source Initiative 문서 첫 문장 'Open source doesn’t just mean access to the source code.' · 라이선스: 남이 그 코드로 할 수 있는 일과 할 수 없는 일을 적은 글. GitHub 문서는 라이선스가 없으면 기본 저작권법이 적용된다고 적음 · stars: 즐겨찾기와 고마움 표시(GitHub 문서). 이 교재는 관심의 크기로 읽기를 권함 · 마지막 업데이트: Releases 칸의 최신 릴리스 날짜와 파일 목록 위 줄의 commit 날짜를 함께 읽기. Whisper 저장소는 릴리스 v20250625(2025년 6월 26일), commit 기록은 두 달 전 · 이슈: 오류 신고와 기능 요청을 적는 글. n8n 저장소 Issues 탭 330, Whisper 저장소는 Issues 탭이 보이지 않음(GitHub 문서: 기능을 끌 수 있음) · 흔한 오해 세 가지: stars 208,482개인 n8n 저장소의 Sustainable Use License, stars 수치와 관리 상태, MIT 라이선스의 조건 · 직접 해 보기: 써 보고 싶은 도구의 repo 페이지에서 네 칸 적기 챕터 0:00 학습 목표 0:20 1장 · 오픈소스와 라이선스 1:19 2장 · stars 수치와 마지막 업데이트 2:40 3장 · 이슈 3:24 4장 · 고르기 전 네 칸 3:47 5장 · 흔한 오해 4:49 6장 · 직접 해 보기 5:10 정리 출처 Open Source Initiative, The Open Source Definition opensource.org GitHub Docs, Licensing a repository docs.github.com GitHub Docs, Saving repositories with stars docs.github.com GitHub Docs, About releases docs.github.com GitHub Docs, About issues docs.github.com GitHub Docs, Disabling issues docs.github.com choosealicense.com, MIT License choosealicense.com openai/whisper openai/whisper n8n-io/n8n LICENSE.md github.com/n8n-io/n8n/blob/master/LICENSE.md AI 교과서 전체 재생목록 → youtube.com DR Builds AI 채널은 AI 공부를 시작한 분을 위한 AI 교과서를 매일 20:00에 한 편씩 올립니다. #AI교과서 #깃허브 #오픈소스 #바이브코딩 #AI입문
Rentabilidad del arriendo en Rionegro por estrato: en cuál rinde más comprar para arrendar En Rionegro, comprar un apartamento para arrendarlo rinde más en estrato 3 (6,6 % bruto al año, 15,2 años de arriendo) y menos en estrato 5 (5,8 %). DOI: 10.5281/zenodo.23521319
Show HN: OmniDiff – Opensource, super fast, content aware Git difftool, in Rust View Article | Discussion + AI Summary Summary of HN discussion 🧵👇
Курсоры для базовых стилей. Рабочая группа CSS закрыла давний спор: элементы форм с appearance: base получат cursor: pointer, если на них можно нажать, grab для ползунков и not-allowed для отключённых. #css #html csswg-drafts#14251 t.co
Курсоры для базовых стилей. Рабочая группа CSS закрыла давний спор: элементы форм с appearance: base получат cursor: pointer, если на них можно нажать, grab для ползунков и not-allowed для отключённых. #css #html csswg-drafts#14251
I’m building an open-source MCP toolkit for source-backed STM32 development I’m working on Praxis, an experimental project that gives AI coding assistants access to reviewed hardware knowledge and scoped firmware tools. The problem I’m trying to address is concrete: generated firmware can compile while using an incorrect clock assumption, an unavailable pin route, or the wrong register access semantics. The knowledge layer is local and vectorless. Sources such as vendor headers, reviewed manual imports and schematic netlists are compiled into versioned Markdown bundles. Retrieval uses canonical names, exact pin/net lookup and explicit prerequisite links. It preserves full published documents and tables. Through MCP, an assistant can retrieve that context. Separate services handle source authoring, isolated compilation, ELF validation and engineering reports. Configured F407 hardware workflows also support exact-image-approved flashing and bounded telemetry. Current focus: STM32F407 Discovery workflows, including LED/button, UART loopback, and timer examples. Other STM32 declaration inventories are exploratory. This is not complete MCU support or production qualification. Some important limits: - PDF extraction still requires interpretation and review. - Dependency gates enforce document retrieval; they don’t prove firmware execution order. - Successful compilation does not prove electrical correctness. - Physical observations apply to the exact tested artifact and setup. Repo: Iheb-AG/praxis If you find the approach useful, a GitHub star is appreciated. Feedback, bug reports, and contributions are equally welcome. I’d appreciate technical criticism: what checks would make this useful in an actual embedded workflow? Are the bigger gaps source interpretation, board topology, or repeatable hardware testing?
6 INSANE GitHub Repos for Claude Code & AI Agents These 6 GitHub repos can change how you work with Claude Code and AI coding agents — from better website design and persistent memory to parallel coding workflows and structured security audits. In this video, we explain Hallmark, Superpowers, Claude Mem, Orca, Cloudflare's Security Audit Skill and TradingAgents. You'll see what each project is designed to do, who it is for, and the installation or setup instructions shown in its documentation. This is a research-based overview of AI tools and GitHub projects, not a hands-on benchmark. TradingAgents is a standalone multi-agent research framework, rather than a Claude Code plugin. 🔎 What you'll discover: • Hallmark — a design skill for more intentional AI-generated websites. • Superpowers — a workflow for brainstorming, planning, test-driven development and review. • Claude Mem — persistent context and searchable memory across agent sessions. • Orca — run coding agents side by side in isolated Git worktrees. • Cloudflare Security Audit Skill — a structured workflow for investigating and validating security findings. • TradingAgents — a multi-agent framework for financial research and simulated trading workflows. 👇 Timestamps — Jump to what matters to you: 0:00 — Why AI-Generated Websites Look the Same 0:48 — #1 Hallmark: Better Design for AI Coding Agents 2:38 — #2 Superpowers: Plan, Build, Test & Review 4:12 — #3 Claude Mem: Persistent Memory Across Sessions 5:55 — #4 Orca: Run Coding Agents in Parallel 7:12 — #5 Cloudflare Security Audit Skill 8:41 — #6 TradingAgents: Multi-Agent Trading Research 9:38 — Final Takeaways & How This Video Was Researched Which repo would you try first — Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill or TradingAgents? Tell us in the comments. Subscribe to NextGen AI for AI tools, GitHub projects, Claude Code workflows and practical AI tutorials. WhatsApp Channel: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 Business: Nextgentech811@gmail.com Accuracy note: This video is based on project READMEs and public pages. We did not install or benchmark these tools. Performance and token-savings figures discussed are the projects' own claims, not independently measured results. Trending positions and star counts change over time; Orca's weekly star figure was not independently verified. Persistent memory does not guarantee perfect recall. Some workflows require paid model access, subscriptions or API usage. This video is for educational purposes only and is not financial, investment or trading advice. TradingAgents is a research framework and does not guarantee profits. Features, setup commands, compatibility, licenses and costs may change; check each project's current documentation. Screenshots and clips are credited to their original creators. #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 Sources & Resources: Hallmark: nutlope/hallmark usehallmark.com Superpowers: obra/superpowers Claude Mem: thedotmack/claude-mem Orca: stablyai/orca onorca.dev Cloudflare Security Audit Skill: cloudflare/security-audit-skill blog.cloudflare.com TradingAgents: TauricResearch/TradingAgents arxiv.org GitHub Trending — Weekly: github.com/trending [de-DE] 6 KRASSE GitHub-Repos für Claude Code & KI-Agenten Diese 6 GitHub-Repos können Ihre Arbeit mit Claude Code und KI-Codierungsagenten revolutionieren – von optimiertem Webdesign und persistentem Speicher bis hin zu parallelen Codierungs-Workflows und strukturierten Sicherheitsaudits. In diesem Video erklären wir Hallmark, Superpowers, Claude Mem, Orca, Cloudflares Security Audit Skill und TradingAgents. Sie erfahren, wofür jedes Projekt konzipiert wurde, für wen es geeignet ist und welche Installations- bzw. Einrichtungsanweisungen in der jeweiligen Dokumentation zu finden sind. Dies ist ein forschungsbasierter Überblick über KI-Tools und GitHub-Projekte, kein praktischer Vergleich. TradingAgents ist ein eigenständiges Multi-Agenten-Forschungsframework und kein Claude-Code-Plugin. 🔎 Was Sie entdecken werden: • Hallmark – eine Design-Skill für zielgerichtetere KI-generierte Websites. • Superpowers – ein Workflow für Brainstorming, Planung, testgetriebene Entwicklung und Überprüfung. • Claude Mem – persistenter Kontext und durchsuchbarer Speicher über Agentensitzungen hinweg. • Orca – ermöglicht das parallele Ausführen von Codierungsagenten in isolierten Git-Arbeitsumgebungen. • Cloudflare Security Audit Skill – ein strukturierter Workflow zur Untersuchung und Validierung von Sicherheitsbefunden. • TradingAgents – ein Multiagenten-Framework für Finanzanalysen und simulierte Handelsabläufe. 👇 Zeitstempel – Springen Sie direkt zu dem, was Sie interessiert: 0:00 – Warum KI-generierte Websites gleich aussehen 0:48 – #1 Hallmark: Besseres Design für KI-Codierungsagenten 2:38 – #2 Superpowers: Planen, Entwickeln, Testen & Überprüfen 4:12 – #3 Claude Mem: Persistenter Speicher über Sitzungen hinweg 5:55 – #4 Orca: Coding-Agenten parallel ausführen 7:12 – #5 Cloudflare Security Audit Skill 8:41 – #6 TradingAgents: Multi-Agent-Trading-Recherche 9:38 – Fazit & Recherchegrundlagen für dieses Video Welches Repository würden Sie zuerst ausprobieren – Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill oder TradingAgents? Teilen Sie es uns in den Kommentaren mit. Abonnieren Sie NextGen AI für KI-Tools, GitHub-Projekte, Claude-Code-Workflows und praktische KI-Tutorials. WhatsApp-Kanal: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 Geschäftlich: Nextgentech811@gmail.com Hinweis zur Genauigkeit: Dieses Video basiert auf den Projekt-READMEs und öffentlichen Seiten. Wir haben diese Tools weder installiert noch Benchmarks durchgeführt. Die genannten Leistungs- und Token-Einsparungszahlen sind Angaben der Projekte selbst und keine unabhängigen Messergebnisse. Trendpositionen und Sterneanzahlen ändern sich im Laufe der Zeit; die wöchentliche Sterneanzahl von Orca wurde nicht unabhängig überprüft. Die Speicherung im Speicher garantiert keine perfekte Wiedergabe. Einige Workflows erfordern kostenpflichtigen Modellzugriff, Abonnements oder die Nutzung der API. Dieses Video dient ausschließlich Bildungszwecken und stellt keine Finanz-, Anlage- oder Handelsberatung dar. TradingAgents ist ein Forschungsframework und garantiert keine Gewinne. Funktionen, Einrichtungsbefehle, Kompatibilität, Lizenzen und Kosten können sich ändern; bitte prüfen Sie die aktuelle Dokumentation des jeweiligen Projekts. Screenshots und Videoclips werden ihren jeweiligen Urhebern zugeordnet. ... #ClaudeCode #GitHub #KIAgenten #KI-Programmierung #KIDerNächstenGeneration 📌 Quellen & Ressourcen: Hallmark: nutlope/hallmark usehallmark.com Superpowers: obra/superpowers Claude Mem: thedotmack/claude-mem Orca: stablyai/orca onorca.dev Cloudflare Security Audit Skill: cloudflare/security-audit-skill blog.cloudflare.com TradingAgents: TauricResearch/TradingAgents arxiv.org GitHub Trends — Wöchentlich: github.com/trending [es-US] 6 Repositorios de GitHub INCREÍBLES para Claude Code y agentes de IA Estos 6 repositorios de GitHub pueden transformar tu forma de trabajar con Claude Code y agentes de codificación de IA, desde un mejor diseño web y memoria persistente hasta flujos de trabajo de codificación paralela y auditorías de seguridad estructuradas. En este video, explicamos Hallmark, Superpowers, Claude Mem, Orca, la habilidad de auditoría de seguridad de Cloudflare y TradingAgents. Verás para qué está diseñado cada proyecto, a quién va dirigido y las instrucciones de instalación o configuración que se muestran en su documentación. Esta es una descripción general de herramientas de IA y proyectos de GitHub basada en investigación, no una prueba comparativa práctica. TradingAgents es un marco de investigación multiagente independiente, no un complemento de Claude Code. 🔎 Lo que descubrirás: • Hallmark: una habilidad de diseño para sitios web generados por IA con mayor intencionalidad. • Superpowers: un flujo de trabajo para la lluvia de ideas, la planificación, el desarrollo guiado por pruebas y la revisión. • Claude Mem: contexto persistente y memoria con capacidad de búsqueda entre sesiones de agentes. • Orca: ejecuta agentes de codificación en paralelo en entornos Git aislados. • Cloudflare Security Audit Skill: un flujo de trabajo estructurado para investigar y validar hallazgos de seguridad. • TradingAgents: un marco multiagente para la investigación financiera y la simulación de flujos de trabajo de trading. 👇 Marcas de tiempo: Ve directamente a lo que te interesa: 0:00 — ¿Por qué los sitios web generados por IA se ven iguales? 0:48 — #1 Hallmark: Mejor diseño para agentes de codificación de IA 2:38 — #2 Superpowers: Planificar, construir, probar y revisar 4:12 — #3 Claude Mem: Memoria persistente entre sesiones 5:55 — #4 Orca: Ejecutar agentes de codificación en paralelo 7:12 — #5 Cloudflare Security Audit Skill 8:41 — #6 TradingAgents: Investigación de trading multiagente 9:38 — Conclusiones finales y cómo se realizó este video ¿Qué repositorio probarías primero: Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill o TradingAgents? Cuéntanos en los comentarios. Suscríbete a NextGen AI para obtener herramientas de IA, proyectos de GitHub, flujos de trabajo de Claude Code y tutoriales prácticos de IA. Canal de WhatsApp: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 Contacto comercial: Nextgentech811@gmail.com Nota sobre la precisión: Este video se basa en los archivos README y las páginas públicas de los proyectos. No instalamos ni realizamos pruebas de rendimiento de estas herramientas. Las cifras de rendimiento y ahorro de tokens que se mencionan son afirmaciones de los propios proyectos, no resultados medidos de forma independiente. Las posiciones en tendencia y el número de estrellas cambian con el tiempo; la cifra semanal de estrellas de Orca no se verificó de forma independiente. La memoria persistente no garantiza una recuperación perfecta. Algunos flujos de trabajo requieren acceso a modelos de pago, suscripciones o el uso de la API. Este video tiene fines educativos únicamente y no constituye asesoramiento financiero, de inversión ni de trading. TradingAgents es un marco de investigación y no garantiza ganancias. Las funciones, los comandos de configuración, la compatibilidad, las licencias y los costos pueden cambiar; consulte la documentación actualizada de cada proyecto. Las capturas de pantalla y los clips pertenecen a sus creadores originales. #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 Fuentes y recursos: Hallmark: nutlope/hallmark usehallmark.com Superpowers: obra/superpowers Claude Mem: thedotmack/claude-mem Orca: stablyai/orca onorca.dev Auditoría de seguridad de Cloudflare Habilidad: cloudflare/security-audit-skill blog.cloudflare.com TradingAgents: TauricResearch/TradingAgents arxiv.org Tendencias semanales de GitHub: github.com/trending [fr-FR] 6 dépôts GitHub DINGUES pour Claude Code & les agents IA Ces 6 dépôts GitHub peuvent transformer votre façon de travailler avec Claude Code et les agents de codage IA : de la conception web optimisée et la mémoire persistante aux flux de travail de codage parallèle et aux audits de sécurité structurés. Dans cette vidéo, nous présentons Hallmark, Superpowers, Claude Mem, Orca, la compétence d'audit de sécurité de Cloudflare et TradingAgents. Vous découvrirez la finalité de chaque projet, son public cible et les instructions d'installation ou de configuration fournies dans sa documentation. Il s'agit d'une présentation théorique des outils d'IA et des projets GitHub, et non d'un test pratique. TradingAgents est un framework de recherche multi-agents autonome, et non un plugin Claude Code. 🔎 Ce que vous découvrirez : • Hallmark : une compétence de conception pour des sites web générés par IA plus pertinents. • Superpowers : un flux de travail pour le brainstorming, la planification, le développement piloté par les tests et la revue de code. • Claude Mem : un contexte persistant et une mémoire consultable entre les sessions des agents. • Orca — Exécution simultanée d'agents de codage dans des environnements Git isolés. • Cloudflare Security Audit Skill — Un flux de travail structuré pour l'analyse et la validation des failles de sécurité. • TradingAgents — Un framework multi-agents pour la recherche financière et les simulations de trading. 👇 Horodatage — Accédez directement à la partie qui vous intéresse : 0:00 — Pourquoi les sites web générés par l’IA se ressemblent-ils ? 0:48 — N° 1 Hallmark : Une meilleure conception pour les agents de codage IA 2:38 — N° 2 Superpowers : Planifier, développer, tester et évaluer 4:12 — N° 3 Claude Mem : Mémoire persistante entre les sessions 5:55 — N° 4 Orca : Exécuter des agents de codage en parallèle 7:12 — N° 5 Cloudflare Security Audit Skill 8:41 — N° 6 TradingAgents : Recherche sur le trading multi-agents 9:38 — Conclusion et méthodologie de recherche Quel dépôt testeriez-vous en premier : Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill ou TradingAgents ? Dites-le-nous en commentaire. Abonnez-vous à NextGen AI pour découvrir des outils d’IA, des projets GitHub, des workflows Claude Code et des tutoriels pratiques sur l’IA. Chaîne WhatsApp : whatsapp.com 🌐 Instagram : instagram.com 📘 Facebook : facebook.com 🎵 TikTok : tiktok.com 📩 Contact professionnel : Nextgentech811@gmail.com Note sur l’exactitude des informations : Cette vidéo est basée sur les fichiers README des projets et les pages publiques. Nous n’avons ni installé ni évalué ces outils. Les chiffres de performance et d’économie de jetons mentionnés sont fournis par les projets eux-mêmes et ne font pas l’objet de mesures indépendantes. Les positions et le nombre d’étoiles évoluent avec le temps ; le nombre d’étoiles hebdomadaire d’Orca n’a pas été vérifié indépendamment. La mémoire persistante ne garantit pas un rappel parfait. Certains flux de travail nécessitent un accès payant aux modèles, des abonnements ou l’utilisation d’une API. Cette vidéo est uniquement destinée à des fins éducatives et ne constitue en aucun cas un conseil financier, d'investissement ou de trading. TradingAgents est un outil de recherche et ne garantit aucun profit. Les fonctionnalités, les commandes d'installation, la compatibilité, les licences et les coûts sont susceptibles d'évoluer ; veuillez consulter la documentation à jour de chaque projet. Les captures d'écran et les extraits vidéo appartiennent à leurs auteurs respectifs. #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 Sources et ressources : Hallmark : nutlope/hallmark usehallmark.com Superpowers : obra/superpowers Claude Mem : thedotmack/claude-mem Orca : stablyai/orca onorca.dev Audit de sécurité Cloudflare Compétences : cloudflare/security-audit-skill blog.cloudflare.com Agents de trading : TauricResearch/TradingAgents arxiv.org Tendances GitHub (hebdomadaires) : github.com/trending [hi] Claude Code और AI एजेंट्स के लिए 6 कमाल के GitHub रिपॉजिटरीज़ ये 6 GitHub रिपॉजिटरी क्लाउड कोड और AI कोडिंग एजेंट के साथ काम करने के आपके तरीके को बदल सकती हैं — बेहतर वेबसाइट डिज़ाइन और स्थायी मेमोरी से लेकर पैरेलल कोडिंग वर्कफ़्लो और संरचित सुरक्षा ऑडिट तक। इस वीडियो में, हम हॉलमार्क, सुपरपावर्स, क्लाउड मेमोरी, ओर्का, क्लाउडफ्लेयर के सिक्योरिटी ऑडिट स्किल और ट्रेडिंग एजेंट्स के बारे में बताते हैं। आप देखेंगे कि प्रत्येक प्रोजेक्ट किस लिए डिज़ाइन किया गया है, यह किसके लिए है, और इसके दस्तावेज़ में दिए गए इंस्टॉलेशन या सेटअप निर्देश। यह AI टूल्स और GitHub प्रोजेक्ट्स का शोध-आधारित अवलोकन है, न कि कोई व्यावहारिक बेंचमार्क। ट्रेडिंग एजेंट्स एक स्टैंडअलोन मल्टी-एजेंट रिसर्च फ्रेमवर्क है, न कि क्लाउड कोड प्लगइन। 🔎 आप क्या जानेंगे: • हॉलमार्क — अधिक उद्देश्यपूर्ण AI-जनरेटेड वेबसाइटों के लिए एक डिज़ाइन स्किल। • सुपरपावर्स — ब्रेनस्टॉर्मिंग, प्लानिंग, टेस्ट-ड्रिवन डेवलपमेंट और रिव्यू के लिए एक वर्कफ़्लो। • क्लाउड मेमोरी — एजेंट सेशन में स्थायी संदर्भ और खोजने योग्य मेमोरी। • ओर्का — अलग-अलग गिट वर्कट्री में कोडिंग एजेंटों को साथ-साथ चलाएं। • क्लाउडफ्लेयर सिक्योरिटी ऑडिट स्किल — सुरक्षा संबंधी निष्कर्षों की जांच और सत्यापन के लिए एक संरचित कार्यप्रवाह। • ट्रेडिंग एजेंट्स — वित्तीय अनुसंधान और सिम्युलेटेड ट्रेडिंग कार्यप्रवाह के लिए एक मल्टी-एजेंट फ्रेमवर्क। 👇 समय-सीमाएँ — सीधे उस भाग पर जाएँ जो आपके लिए महत्वपूर्ण है: 0:00 — AI द्वारा निर्मित वेबसाइटें एक जैसी क्यों दिखती हैं? 0:48 — #1 हॉलमार्क: AI कोडिंग एजेंटों के लिए बेहतर डिज़ाइन 2:38 — #2 सुपरपावर्स: योजना बनाएँ, निर्माण करें, परीक्षण करें और समीक्षा करें 4:12 — #3 क्लाउड मेम: सत्रों के दौरान स्थायी मेमोरी 5:55 — #4 ओर्का: कोडिंग एजेंटों को समानांतर रूप से चलाएँ 7:12 — #5 क्लाउडफ्लेयर सुरक्षा ऑडिट कौशल 8:41 — #6 ट्रेडिंग एजेंट: मल्टी-एजेंट ट्रेडिंग अनुसंधान 9:38 — मुख्य निष्कर्ष और इस वीडियो के लिए शोध कैसे किया गया आप सबसे पहले कौन सा रिपॉजिटरी आज़माना चाहेंगे — हॉलमार्क, सुपरपावर्स, क्लाउड मेम, ओर्का, सुरक्षा ऑडिट कौशल या ट्रेडिंग एजेंट? हमें टिप्पणियों में बताएँ। AI टूल्स, GitHub प्रोजेक्ट्स, क्लाउड कोड वर्कफ़्लो और व्यावहारिक AI ट्यूटोरियल के लिए NextGen AI को सब्सक्राइब करें। WhatsApp चैनल: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 व्यावसायिक संपर्क: Nextgentech811@gmail.com सटीकता नोट: यह वीडियो प्रोजेक्ट के README और सार्वजनिक पृष्ठों पर आधारित है। हमने इन टूल्स को इंस्टॉल या बेंचमार्क नहीं किया है। प्रदर्शन और टोकन बचत के आंकड़े प्रोजेक्ट के स्वयं के दावे हैं, स्वतंत्र रूप से मापे गए परिणाम नहीं। ट्रेंडिंग पोजीशन और स्टार काउंट समय के साथ बदलते रहते हैं; Orca के साप्ताहिक स्टार आंकड़े का स्वतंत्र रूप से सत्यापन नहीं किया गया है। स्थायी मेमोरी सटीक रिकॉल की गारंटी नहीं देती है। कुछ वर्कफ़्लो के लिए सशुल्क मॉडल एक्सेस, सब्सक्रिप्शन या API उपयोग की आवश्यकता होती है। यह वीडियो केवल शैक्षिक उद्देश्यों के लिए है और इसे वित्तीय, निवेश या ट्रेडिंग सलाह के रूप में नहीं लिया जाना चाहिए। ट्रेडिंग एजेंट्स एक शोध ढांचा है और लाभ की गारंटी नहीं देता है। विशेषताएं, सेटअप कमांड, अनुकूलता, लाइसेंस और लागत में बदलाव हो सकते हैं; कृपया प्रत्येक प्रोजेक्ट के वर्तमान दस्तावेज़ देखें। स्क्रीनशॉट और वीडियो क्लिप उनके मूल रचनाकारों को श्रेय दिए जाते हैं। #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 स्रोत और संसाधन: हॉलमार्क: nutlope/hallmark usehallmark.com सुपरपावर्स: obra/superpowers क्लाउड मेम: thedotmack/claude-mem ओर्का: stablyai/orca onorca.dev क्लाउडफ्लेयर सुरक्षा ऑडिट कौशल: cloudflare/security-audit-skill blog.cloudflare.com ट्रेडिंग एजेंट: TauricResearch/TradingAgents arxiv.org GitHub ट्रेंडिंग — साप्ताहिक: github.com/trending [id] 6 Repo GitHub GILA untuk Claude Code & AI Agent Enam repositori GitHub ini dapat mengubah cara Anda bekerja dengan Claude Code dan agen pengkodean AI — mulai dari desain situs web yang lebih baik dan memori persisten hingga alur kerja pengkodean paralel dan audit keamanan terstruktur. Dalam video ini, kami menjelaskan Hallmark, Superpowers, Claude Mem, Orca, Cloudflare's Security Audit Skill, dan TradingAgents. Anda akan melihat apa yang dirancang untuk setiap proyek, untuk siapa proyek tersebut, dan petunjuk instalasi atau pengaturan yang ditunjukkan dalam dokumentasinya. Ini adalah tinjauan berbasis riset tentang alat AI dan proyek GitHub, bukan tolok ukur langsung. TradingAgents adalah kerangka kerja riset multi-agen mandiri, bukan plugin Claude Code. 🔎 Apa yang akan Anda temukan: • Hallmark — keterampilan desain untuk situs web yang dihasilkan AI yang lebih terencana. • Superpowers — alur kerja untuk brainstorming, perencanaan, pengembangan berbasis pengujian, dan peninjauan. • Claude Mem — konteks persisten dan memori yang dapat dicari di seluruh sesi agen. • Orca — menjalankan agen pengkodean berdampingan dalam pohon kerja Git yang terisolasi. • Cloudflare Security Audit Skill — alur kerja terstruktur untuk menyelidiki dan memvalidasi temuan keamanan. • TradingAgents — kerangka kerja multi-agen untuk riset keuangan dan alur kerja perdagangan simulasi. 👇 Cap waktu — Langsung ke bagian yang penting bagi Anda: 0:00 — Mengapa Situs Web yang Dihasilkan AI Terlihat Sama 0:48 — #1 Hallmark: Desain yang Lebih Baik untuk Agen Pengkodean AI 2:38 — #2 Superpowers: Rencanakan, Bangun, Uji & Tinjau 4:12 — #3 Claude Mem: Memori Persisten Lintas Sesi 5:55 — #4 Orca: Jalankan Agen Pengkodean Secara Paralel 7:12 — #5 Cloudflare Security Audit Skill 8:41 — #6 TradingAgents: Riset Perdagangan Multi-Agen 9:38 — Kesimpulan Akhir & Bagaimana Video Ini Diriset Repositori mana yang akan Anda coba terlebih dahulu — Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill, atau TradingAgents? Beri tahu kami di kolom komentar. Berlangganan NextGen AI untuk alat AI, proyek GitHub, alur kerja Claude Code, dan tutorial AI praktis. Saluran WhatsApp: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 Bisnis: Nextgentech811@gmail.com Catatan Akurasi: Video ini didasarkan pada README proyek dan halaman publik. Kami tidak menginstal atau melakukan benchmark alat-alat ini. Angka kinerja dan penghematan token yang dibahas adalah klaim proyek sendiri, bukan hasil yang diukur secara independen. Posisi tren dan jumlah bintang berubah seiring waktu; angka bintang mingguan Orca tidak diverifikasi secara independen. Memori persisten tidak menjamin ingatan yang sempurna. Beberapa alur kerja memerlukan akses model berbayar, langganan, atau penggunaan API. Video ini hanya untuk tujuan pendidikan dan bukan merupakan nasihat keuangan, investasi, atau perdagangan. TradingAgents adalah kerangka kerja riset dan tidak menjamin keuntungan. Fitur, perintah pengaturan, kompatibilitas, lisensi, dan biaya dapat berubah; periksa dokumentasi terkini setiap proyek. Tangkapan layar dan klip dikreditkan kepada pencipta aslinya. #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 Sumber & Referensi: Hallmark: nutlope/hallmark usehallmark.com Superpowers: obra/superpowers Claude Mem: thedotmack/claude-mem Orca: stablyai/orca onorca.dev Cloudflare Security Audit Skill: cloudflare/security-audit-skill blog.cloudflare.com TradingAgents: TauricResearch/TradingAgents arxiv.org GitHub Trending — Mingguan: github.com/trending [it] 6 incredibili repository GitHub per Claude Code e agenti IA Questi 6 repository GitHub possono rivoluzionare il tuo modo di lavorare con Claude Code e gli agenti di programmazione basati sull'IA, offrendo una migliore progettazione di siti web, memoria persistente, flussi di lavoro di programmazione parallela e audit di sicurezza strutturati. In questo video, spieghiamo Hallmark, Superpowers, Claude Mem, Orca, la skill Security Audit di Cloudflare e TradingAgents. Scoprirai a cosa serve ciascun progetto, a chi è destinato e le istruzioni di installazione o configurazione presenti nella relativa documentazione. Questa è una panoramica basata sulla ricerca di strumenti di IA e progetti GitHub, non un benchmark pratico. TradingAgents è un framework di ricerca multi-agente autonomo, non un plugin per Claude Code. 🔎 Cosa scoprirai: • Hallmark: una skill di progettazione per siti web generati dall'IA più mirati. • Superpowers: un flusso di lavoro per brainstorming, pianificazione, sviluppo guidato dai test e revisione. • Claude Mem: contesto persistente e memoria ricercabile tra le sessioni degli agenti. • Orca: consente di eseguire agenti di programmazione in parallelo in directory di lavoro Git isolate. • Cloudflare Security Audit Skill: un flusso di lavoro strutturato per l'analisi e la convalida delle vulnerabilità di sicurezza. • TradingAgents: un framework multi-agente per la ricerca finanziaria e la simulazione di flussi di lavoro di trading. 👇 Indice dei contenuti — Vai direttamente alla sezione che ti interessa: 0:00 — Perché i siti web generati dall'IA sembrano tutti uguali 0:48 — #1 Hallmark: Design migliore per gli agenti di programmazione IA 2:38 — #2 Superpowers: Pianifica, costruisci, testa e rivedi 4:12 — #3 Claude Mem: Memoria persistente tra sessioni 5:55 — #4 Orca: Esegui agenti di programmazione in parallelo 7:12 — #5 Cloudflare Security Audit Skill 8:41 — #6 TradingAgents: Ricerca sul trading multi-agente 9:38 — Conclusioni e come è stata realizzata la ricerca per questo video Quale repository proveresti per primo: Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill o TradingAgents? Diccelo nei commenti. Iscriviti a NextGen AI per strumenti IA, progetti GitHub, flussi di lavoro Claude Code e tutorial pratici sull'IA. Canale WhatsApp: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 Contatti commerciali: Nextgentech811@gmail.com Nota sull'accuratezza: questo video si basa sui file README dei progetti e sulle pagine pubbliche. Non abbiamo installato né testato questi strumenti. Le cifre relative alle prestazioni e al risparmio di token discusse sono dichiarazioni dei progetti stessi e non risultati misurati in modo indipendente. Le posizioni in classifica e il numero di stelle cambiano nel tempo; il dato settimanale di stelle di Orca non è stato verificato in modo indipendente. La memoria persistente non garantisce un ricordo perfetto. Alcuni flussi di lavoro richiedono l'accesso a modelli a pagamento, abbonamenti o l'utilizzo di API. Questo video è a scopo puramente didattico e non costituisce consulenza finanziaria, di investimento o di trading. TradingAgents è un framework di ricerca e non garantisce profitti. Funzionalità, comandi di configurazione, compatibilità, licenze e costi possono subire variazioni; si prega di consultare la documentazione aggiornata di ciascun progetto. Gli screenshot e le clip sono attribuiti ai rispettivi creatori. #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 Fonti e risorse: Hallmark: nutlope/hallmark usehallmark.com Superpowers: obra/superpowers Claude Mem: thedotmack/claude-mem Orca: stablyai/orca onorca.dev Cloudflare Security Audit Skill: cloudflare/security-audit-skill blog.cloudflare.com TradingAgents: TauricResearch/TradingAgents arxiv.org GitHub Tendenze settimanali: github.com/trending [pt-BR] 6 Repositórios INSANOS do GitHub para Claude Code e Agentes de IA Estes 6 repositórios do GitHub podem mudar a forma como você trabalha com o Claude Code e agentes de IA para codificação — desde um design de sites aprimorado e memória persistente até fluxos de trabalho de codificação paralelos e auditorias de segurança estruturadas. Neste vídeo, explicamos o Hallmark, o Superpowers, o Claude Mem, o Orca, a Skill de Auditoria de Segurança da Cloudflare e o TradingAgents. Você verá para que cada projeto foi desenvolvido, para quem ele é indicado e as instruções de instalação ou configuração presentes em sua documentação. Esta é uma visão geral baseada em pesquisa sobre ferramentas de IA e projetos do GitHub, não um benchmark prático. O TradingAgents é uma estrutura de pesquisa multiagente independente, e não um plugin do Claude Code. 🔎 O que você descobrirá: • Hallmark — uma skill de design para sites gerados por IA mais intencionais. • Superpowers — um fluxo de trabalho para brainstorming, planejamento, desenvolvimento orientado a testes e revisão. • Claude Mem — contexto persistente e memória pesquisável entre sessões de agentes. • Orca — execute agentes de codificação lado a lado em árvores de trabalho Git isoladas. • Cloudflare Security Audit Skill — um fluxo de trabalho estruturado para investigar e validar descobertas de segurança. • TradingAgents — uma estrutura multiagente para pesquisa financeira e fluxos de trabalho de negociação simulada. 👇 Marcadores de tempo — Vá direto ao que interessa: 0:00 — Por que os sites gerados por IA são tão parecidos? 0:48 — #1 Hallmark: Melhor design para agentes de codificação de IA 2:38 — #2 Superpowers: Planejar, construir, testar e revisar 4:12 — #3 Claude Mem: Memória persistente entre sessões 5:55 — #4 Orca: Executar agentes de codificação em paralelo 7:12 — #5 Cloudflare Security Audit Skill 8:41 — #6 TradingAgents: Pesquisa de negociação multiagente 9:38 — Conclusões finais e como este vídeo foi pesquisado Qual repositório você testaria primeiro — Hallmark, Superpowers, Claude Mem, Orca, Security Audit Skill ou TradingAgents? Conte para nós nos comentários. Inscreva-se no NextGen AI para receber ferramentas de IA, projetos do GitHub, fluxos de trabalho do Claude Code e tutoriais práticos de IA. Canal do WhatsApp: whatsapp.com 🌐 Instagram: instagram.com 📘 Facebook: facebook.com 🎵 TikTok: tiktok.com 📩 Contato comercial: Nextgentech811@gmail.com Nota sobre precisão: Este vídeo é baseado nos arquivos README e páginas públicas dos projetos. Não instalamos nem testamos essas ferramentas. Os dados de desempenho e economia de tokens mencionados são afirmações dos próprios projetos, não resultados medidos independentemente. As posições em alta e a contagem de estrelas mudam com o tempo; o número de estrelas semanais do Orca não foi verificado independentemente. A memória persistente não garante recall perfeito. Alguns fluxos de trabalho exigem acesso pago ao modelo, assinaturas ou uso da API. Este vídeo tem fins exclusivamente educativos e não constitui aconselhamento financeiro, de investimento ou de negociação. O TradingAgents é uma plataforma de pesquisa e não garante lucros. Funcionalidades, comandos de configuração, compatibilidade, licenças e custos podem sofrer alterações; consulte a documentação atual de cada projeto. As capturas de tela e os vídeos pertencem aos seus criadores originais. #ClaudeCode #GitHub #AIAgents #AICoding #NextGenAI 📌 Fontes e Recursos: Hallmark: nutlope/hallmark usehallmark.com Superpowers: obra/superpowers Claude Mem: thedotmack/claude-mem Orca: stablyai/orca onorca.dev Auditoria de Segurança da Cloudflare Skill: cloudflare/security-audit-skill blog.cloudflare.com TradingAgents: TauricResearch/TradingAgents arxiv.org Tendências do GitHub — Semanal: github.com/trending
GLM 5.3 Flash on a 12GB GPU: Project Maya Setup & Tests GLM 5.3 Flash is a 321-billion-parameter model. Project Maya, a brand-new open-source engine, splits the model across the graphics card, RAM and SSD, and tunes itself to your PC. I set it up, ran the full test set, and compared it with DeepSeek V4.1 Flash on Strata on the exact same graphics card. Every number in this video comes from a real run on real hardware. Project Maya is a new, small project — take this as an exciting first look, not the final word. What you'll see: ✅ One setup script — PC check, model choice, download, self-tuning ✅ How a 321B mixture-of-experts model fits: experts on GPU → RAM → SSD, quantization, the 4 Maya builds ✅ Writing vs reading ✅ Real results: AIME 4/4, long-document recall 5/5, vision, 4 languages, 5 apps built from one prompt, code with hidden tests ✅ Same PC, fair fight: GLM on Maya vs DeepSeek V4.1 Flash on Strata 🔗 Commands, test scripts, outputs and raw logs: network-tocoder/GLM-5.3-Flash-321B-on-a-12GB-GPU-Project-Maya 🔗 Project Maya: mw00/project-maya 🔗 Model: huggingface.co 🔗 Maya builds: huggingface.co 🔗 Last video (DeepSeek V4.1 Flash 763B on 12GB, Strata): network-tocoder/DeepSeek-V4.1-Flash-763B-on-a-12GB-GPU-Strata 🖥️ Test setup GPU: RTX 5070 · 12 GB VRAM RAM: 60 GB Storage: NVMe SSD Model: GLM-5.3-Flash #LocalAI #GLM53 #ProjectMaya #LocalLLM #RTX5070 #OpenSourceAI #AI 💜 If these tests are useful, consider becoming a member — your support funds every future test.
사진 한 장을 세트장처럼 걸어 다닌다면? GitHub 별 1.2만 image-blaster, 사진을 3D 세계로 #Shorts 사진 한 장을 영화 세트장처럼 들어가 둘러볼 수 있다면 어떨까요? image-blaster(neilsonnn/image-blaster)는 GitHub 별 약 1만 2천 개(12.3k, 2026.11.01 GitHub API 집계)의 MIT 오픈소스로, 저장소 설명은 'Claude용 이미지-투-월드 스킬 모음'입니다. 스킬은 Claude에게 작업 순서를 알려 주는 설명서 묶음입니다. README에 따르면 Claude 스킬과 World Labs, FAL을 이용해 이미지 한 장에서 3D 환경, 효과음, 3D 모형을 만듭니다. · 기본 결과물: 움직일 물건의 3D 모형(.glb, .obj), 고정된 배경의 가우시안 스플랫(.spz, 둘러볼 수 있는 3D 공간), 반복 배경음과 물건별 물리 효과음(.mp3) · README 소개: 이미지에서 모형까지 갖춘 3D 환경까지 5분 이내 · 빠른 시작: 저장소를 받아 폴더에서 Claude Code를 실행 → World Labs·FAL API 키를 Claude에게 전달 → input/ 폴더에 사진을 넣고 'blast it and confirm each step with me'라고 요청 · 사용 모델(README): World Labs marble-1.1(3D 공간), nano-banana·gpt-image-2(이미지 편집), FAL을 통한 Hunyuan 3D(3D 모형), ElevenLabs SFX(효과음) · 결과물은 Unity·Unreal·Godot, Blender·3DS Max·Maya, Three.js·Electron 등으로 가져가 이어 작업 가능 주의: World Labs와 FAL의 API 키(외부 서비스 이용 열쇠)가 필요합니다. README에는 비용 안내가 없으니 각 서비스 요금을 따로 확인하세요. '5분 이내'는 개발자 소개 문구로 MIRLABS가 직접 재 보지 않았습니다. 처음엔 README 예시처럼 단계마다 확인하며 실행하는 걸 권합니다. README 기반 소개이며 직접 설치해 보지 않았습니다. 화면의 README 캡처는 GitHub 원문(raw README)을 MIRLABS가 렌더링한 것이고, 첫 장면의 사진→3D 그림은 개념 도식(실제 화면 아님)입니다. 출처 neilsonnn/image-blaster raw.githubusercontent.com MIRLABS · AI 도구와 기술을 짧게 #MIRLABS #미르랩스 #imageblaster #3D #가우시안스플랫 #ClaudeCode #WorldLabs #게임개발 #Shorts
🛡️ A small network intrusion detection system in pure Rust. sentinel parses Ethernet, IPv4/6, TCP, UDP, ICMP and ARP zero-copy, matches pcap captures against readable TOML rules and prints colored alerts. Live capture is next. #rustlang #infosec quantumnic/sentinel t.co
This Free Open-Source Repo Syncs Notes Between Devices on the Same Wi-Fi ScepticDope/Minoo #OpenSource t.co
Rentabilidad del arriendo en Pereira por estrato: en cuál rinde más comprar para arrendar En Pereira, comprar un apartamento para arrendarlo rinde más en estrato 4 (7,4 % bruto al año, 13,5 años de arriendo) y menos en estrato 5 (6,6 %). DOI: 10.5281/zenodo.23521223
Cursors for base appearance. CSSWG settled the long-standing debate: form controls with appearance: base get cursor: pointer when clickable, grab on range sliders, and not-allowed when disabled. #css #html csswg-drafts#14251
Redis 8.12 RC2's SEXPIRE command cuts a set membership check from two round trips to one --- title: "Redis 8.12 RC2's SEXPIRE command cuts a set membership check from two round trips to one" published: true published_at: 2026-11-02 13:00 +0200 description: "Redis 8.12 RC2 adds per-member TTLs to sets. A membership check dropped from roughly 585 to 290 microseconds over 2,000 calls, but setting the TTL itself wasn't faster, and converting a small intset cost 15 times more memory." tags: redis, database, performance, devops ai_disclosure_level: ai_assisted --- A Redis set has always had one expiry for the whole key. If you wanted one member of a set to disappear on its own, say a session token in a set of active sessions, or a temporary grant in a set of permissions, you built a side index yourself: a sorted set with the expiry as the score, checked on every read, swept by a cron job. Redis 8.12, currently at release candidate 2, adds that directly to the set type. I built it from source, since there's no Docker image for an RC yet, and measured what changes when the server does this instead of your application code. The new commands are SEXPIRE, SPEXPIRE, SEXPIREAT, SPEXPIREAT to set an expiry on specific members, SPERSIST to clear it, and STTL, SPTTL, SEXPIRETIME, SPEXPIRETIME to read it back. There's no official command reference page for any of them yet; I pieced the syntax together from COMMAND DOCS and the release notes. The headline number The actual value of this feature isn't in setting the expiry. It's in checking membership afterwards. Under the old pattern, "is this member still valid" means SISMEMBER plus a ZSCORE lookup against your own expiry index, compared against the current time client-side. Under the new one, SISMEMBER alone already reflects the TTL, because the server has removed the member once it expired. I populated two 2,000-member sets, gave every member the same TTL, then ran 2,000 validity checks against each pattern, three times: | run | old pattern (SISMEMBER + ZSCORE) | new pattern (SISMEMBER alone) | difference | |---|---|---|---| | 1 | 997.7 ms | 475.8 ms | 52.3% faster | | 2 | 889.0 ms | 444.6 ms | 50.0% faster | | 3 | 994.3 ms | 423.6 ms | 57.4% faster | That's one round trip removed from every single read, and it held steady across all three runs: roughly half the time, every time. On loopback, where a round trip is already cheap, that's still a 50% cut. Over a real network to a managed Redis instance, removing a round trip matters more, not less, though I only measured loopback here and I'm not going to claim a number I didn't run. The number that argues against it Setting the expiry doesn't get any faster. I benchmarked writing the TTL itself, 2,000 members, three runs, old pattern being a plain ZADD into a side index versus the new SEXPIRE call: | run | old pattern (ZADD to a side index) | new pattern (SEXPIRE) | difference | |---|---|---|---| | 1 | 497.2 ms | 506.7 ms | 1.9% slower | | 2 | 430.0 ms | 438.9 ms | 2.1% slower | | 3 | 483.7 ms | 599.7 ms | 24.0% slower | SEXPIRE was never faster to call than a plain ZADD, and in one run it was a quarter slower. That's not surprising once you know what it has to do: find the member inside the set's internal structure, possibly convert the set's whole encoding to accommodate per-member metadata, and schedule it for the active-expiry cycle. A ZADD into an unrelated sorted set does none of that. The saving in this feature is entirely on the read side, where you stop needing the side index at all. If your workload only ever writes expiring members and never re-checks them, this feature costs you a little and gives nothing back. What it costs Nobody mentions this part. A set that has never had a member TTL stores members as compactly as possible: a plain intset for small sets of integers, a listpack for small sets of short strings. The moment any member gets a TTL, the set needs a separate expiry slot for every member, TTL or not, and the encoding changes to carry that. For a listpack-encoded set of short strings, the conversion to listpackex cost 42-75% more memory, worse for smaller sets since the per-entry overhead is proportionally larger: | set size | before | after SEXPIRE on one member | overhead | |---|---|---|---| | 10 members | 70 bytes | 122 bytes | +74.3% | | 50 members | 270 bytes | 402 bytes | +48.9% | | 100 members | 521 bytes | 753 bytes | +44.5% | | 128 members | 689 bytes | 977 bytes | +41.8% | That's expected, if undocumented. What I didn't expect: a small intset, 300 plain integers, has no TTL-aware equivalent at all. Giving one member a TTL doesn't convert it to some integer-specific expiring encoding, it jumps straight to a full hashtable: encoding before: intset, 636 bytes encoding after SEXPIRE on one member: hashtable, 10388 bytes delta: +9752 bytes (+1533.3%) A 15-times memory increase, for giving one out of 300 integers an expiry. And it doesn't come back. I called SPERSIST to remove the TTL again, and the set stayed on listpackex/hashtable rather than reverting to listpack/intset: > SPERSIST t MEMBERS 1 a (integer) 1 > OBJECT ENCODING t "listpackex" If you're going to use this on a set that's mostly stable and rarely touched by expiry, know that touching it once is permanent, memory-wise, until the key is rewritten some other way. What it refuses SEXPIRE won't take a negative TTL, which is a real departure from the key-level EXPIRE command, where a negative or past expiry deletes the key immediately: > SEXPIRE myset -1 MEMBERS 1 x (error) ERR invalid expire time, must be >= 0 If you want to delete a member immediately, you still use SREM; SEXPIRE with 0 works and deletes it on the next expiry pass, but a negative value is simply rejected. It also refuses to run on the wrong type: > SET strkey hello > SEXPIRE strkey 100 MEMBERS 1 x (error) WRONGTYPE Operation against a key holding the wrong kind of value And a nonexistent key or nonexistent member both return -2 from STTL, same as the hash-field TTL commands that came before this. Under concurrency I fired fifty concurrent SEXPIRE calls at the same member from fifty threads, each setting a different TTL. No errors, no corruption, and the final TTL matched whichever call the server happened to execute last, since Redis serialises command execution regardless of how many clients are connected. This isn't really a test of the feature so much as a reminder that the usual Redis guarantee holds here too: nothing about per-member TTLs introduces a new race inside the server. The race that matters is the one on your side, between reading a member and acting on it, and that hasn't changed. A documented claim, checked I couldn't find a documented statement about whether SINTERSTORE and SUNIONSTORE preserve member TTLs on the destination set, so I checked directly. They do: > SADD s1 x y z > SEXPIRE s1 100 MEMBERS 1 x > SINTERSTORE dest s1 > STTL dest MEMBERS 1 x (integer) 100 The TTL on x carried through to dest with the same remaining time. I'd half expected a store operation to rebuild the set fresh and drop per-member metadata the way some operations silently reset a counter or a type. It doesn't. Watching it in production INFO stats has two new counters: expired_subkeys and expired_subkeys_active. INFO keyspace grew a subexpiry field showing how many keys in the database currently have at least one member or field with a TTL. I set a short TTL on two members of a five-member set, never read the key again, and watched the counters change on their own: keyspace right after: subexpiry=1 expired_subkeys_active before: 3 (... 3 second wait, no access to the key at all ...) expired_subkeys_active after: 5 keyspace after: subexpiry=0 That confirms the background active-expiry cycle genuinely sweeps expired set members without anything ever reading the key, which matters if your workload sets TTLs on keys it then rarely touches. Without that active sweep, memory for expired-but-unread members would sit around until something happened to read the key and trigger lazy cleanup. What I got wrong on the way I tried to check whether SINTERSTORE preserves TTLs by running SINTERSTORE dest 1 s1, copying the numkeys argument style from SINTERCARD. The result came back as an empty set every time, with or without a TTL anywhere near it. I nearly wrote that down as "SINTERSTORE silently drops everything when any source member has a TTL", which would have been a dramatic finding and also completely wrong. SINTERSTORE doesn't take a numkeys argument at all; its syntax is just SINTERSTORE destination key [key ...]. My 1 was being read as a literal key named "1", which doesn't exist, so the intersection was always empty, TTL or no TTL. Running the same broken command against a set with zero TTL activity produced the identical empty result, which is what gave it away: a real TTL-related bug wouldn't also break on a completely untouched set. Run it yourself There's no Docker image for 8.12 RC2 yet, so building from source is the fastest path: git clone --depth 1 --branch 8.12-rc2 https://github.com/redis/redis.git cd redis && make -j4 BUILD_TLS=no ./src/redis-server --daemonize yes --save "" --port 6379 The commands behind the numbers above, run against that server: redis-cli SADD myset a b c redis-cli SEXPIRE myset 100 MEMBERS 1 a redis-cli STTL myset MEMBERS 1 a # 100 redis-cli STTL myset MEMBERS 1 b # -1, no TTL set redis-cli OBJECT ENCODING myset # listpackex, once any member has a TTL The two timing benchmarks, with redis-py: import redis, time r = redis.Redis(port=6379) N = 2000 r.sadd('newset', *[f'm{i}' for i in range(N)]) for i in range(N): r.execute_command('SEXPIRE', 'newset', 600, 'MEMBERS', 1, f'm{i}') r.sadd('oldset', *[f'm{i}' for i in range(N)]) now = time.time() for i in range(N): r.zadd('oldset:idx', {f'm{i}': now + 600}) t0 = time.perf_counter() for i in range(N): r.sismember('newset', f'm{i}') print('native:', time.perf_counter() - t0) t0 = time.perf_counter() for i in range(N): r.sismember('oldset', f'm{i}') r.zscore('oldset:idx', f'm{i}') print('side-index:', time.perf_counter() - t0) If you're maintaining a side index purely to expire individual set members, and you check membership more often than you set it, moving to SEXPIRE on 8.12 removes a round trip from every read for roughly double the read throughput, at the cost of a permanent jump in the set's memory footprint and a write path that's no faster than what you had before. Check which of those two costs you more before switching, and don't bother if you're only ever setting expiries and never checking them back.
Cursors for base appearance. CSSWG settled the long-standing debate: form controls with appearance: base get cursor: pointer when clickable, grab on range sliders, and not-allowed when disabled. #css #html csswg-drafts#14251 t.co
BootGPT fits inference code in a 512-byte boot sector, with a separate 11.5 KB int8 character model. It generates synthetic boot logs, not general-purpose answers or real diagnostics. kerlenton/bootgpt #TinyInferenceSystems #QuantizedRuntime t.co
@IlFglososin @Sammiken github.com/SnapdragonGameStudios/snapdragon-gsr/b… вроде етот получше будет
Rentabilidad del arriendo en Medellín por estrato: en cuál rinde más comprar para arrendar En Medellín, comprar un apartamento para arrendarlo rinde más en estrato 2 (8,1 % bruto al año, 12,4 años de arriendo) y menos en estrato 4 (7,4 %). DOI: 10.5281/zenodo.23519067
AI generates mountains of code. How do I understand it at a high level? I use a flow-chart skill. I've listed the skill repo below. You can not only visualize things, you can discuss it with AI cause every element has a unique name in my diagrams. naeemakram/flow-chart-skill
@Jennyliu2728 将eternity4719/HowToLiveBetter%E4%B8%8B%E8%BD%BDpdf%E6%96%87%E4%BB%B6%E5%96%82%E7%BB%99ai%E5%B7%A5%E5%85%B7%EF%BC%8C%E7%84%B6%E5%90%8E%E8%AE%A9%E4%BB%96%E5%88%86%E6%9E%90%E5%87%BA%E7%9F%A5%E8%AF%86%E5%8D%A1%E7%89%87%E3%80%82 多尝试几次,然后逐条出
Rentabilidad del arriendo en Manizales por estrato: en cuál rinde más comprar para arrendar En Manizales, comprar un apartamento para arrendarlo rinde más en estrato 4 (7,2 % bruto al año, 14,0 años de arriendo) y menos en estrato 5 (5,7 %). DOI: 10.5281/zenodo.23521700
Parallax: free, open-source photogrammetry that runs entirely on your Android phone. Looking for testers and reviewers I've been building Parallax, a GPL-3.0 Android app that does the whole photogrammetry pipeline on the phone. You walk around an object, it picks sharp keyframes as you go, and it reconstructs a coloured point cloud locally. No cloud service, and the app has no network permission at all. What's in it - Incremental SfM: SIFT (RootSIFT), five-point RANSAC, PnP registration and a Schur-complement bundle adjustment that also refines the lens (focal length, k1, k2) - Dense: rectified semi-global matching between neighbouring views, cross-checked and fused into one cloud with normals - Metric scale on phones that let apps stream two rear cameras at once, using the factory lens positions as a known baseline. On other phones the model is unscaled - Models come out gravity-up; exposure, white balance and focus are locked during a scan; the torch switches on in dim light and stays fixed while scanning - Export to PLY, or share the whole scan as a ready COLMAP project if you want to mesh or refine it on a desktop Where it stands This is v0.1. The reconstruction core is checked against a synthetic ground-truth scene: camera positions within half a millimetre, rig scale within 0.2 %, median surface error under 1 mm at about 60 cm. The full app flow runs on the Android emulator, which is also where the screenshots come from. What I don't have yet is results from real phones, and that's where I'd love your help. How you can help - Test it. Android 10+, 64-bit ARM. Grab the APK from Releases, scan something with texture, and open an issue with your phone model, the capture modes the app offers (the chips at the top of the camera screen), and whether the scan reconstructed. Attaching the exported ZIP helps a lot. - Review it. If you know SfM/MVS, the core in app/src/main/cpp/core is plain C++17 + OpenCV and builds on a desktop with a test harness. I'd value a critical look. - Contribute. Next on the list: meshing (Poisson), a GPU path for dense matching, and building OpenCV from source so the app can go on F-Droid. Foadsf/parallax
Rentabilidad del arriendo en La Ceja por estrato: en cuál rinde más comprar para arrendar En La Ceja, comprar un apartamento para arrendarlo rinde más en estrato 4 (7,1 % bruto al año, 14,2 años de arriendo) y menos en estrato 5 (6,0 %). DOI: 10.5281/zenodo.23522408