Hey HN, we trained and open-sourced a 1.5B model that predicts your next edits, similar to Cursor. You can download the weights here (<a href="https:&#x2F;&#x2F;huggingface.co&#x2F;sweepai&#x2F;sweep-next-edit-1.5b" rel="nofollow">https:&#x2F;&#x2F;huggingface.co&#x2F;sweepai&#x2F;sweep-next-edit-1.5b</a>) or try it in our JetBrains plugin (<a href="https:&#x2F;&#x2F;plugins.jetbrains.com&#x2F;plugin&#x2F;26860-sweep-ai-autocomplete--coding-agent" rel="nofollow">https:&#x2F;&#x2F;plugins.jetbrains.com&#x2F;plugin&#x2F;26860-sweep-ai-autocomp...</a>).Next-edit autocomplete differs from standard autocomplete by using your recent edits as context when predicting completions. The model is small enough to run locally while outperforming models 4x its size on both speed and accuracy.We tested against Mercury (Inception), Zeta (Zed), and Instinct (Continue) across five benchmarks: next-edit above&#x2F;below cursor, tab-to-jump for distant changes, standard FIM, and noisiness. We found exact-match accuracy correlates best with real usability because code is fairly precise and the solution space is small.Prompt format turned out to matter more than we expected. We ran a genetic algorithm over 30+ diff formats and found simple `original`&#x2F;`updated` blocks beat unified diffs. The verbose format is just easier for smaller models to understand.Training was SFT on ~100k examples from permissively-licensed repos (4hrs on 8xH100), then RL for 2000 steps with tree-sitter parse checking and size regularization. The RL step fixes edge cases SFT can’t like, generating code that doesn’t parse or overly verbose outputs.We&#x27;re open-sourcing the weights so the community can build fast, privacy-preserving autocomplete for any editor. If you&#x27;re building for VSCode, Neovim, or something else, we&#x27;d love to see what you make with it!

Show HN: Sweep, Open-weights 1.5B model for next-edit autocomplete

I built this for myself while debugging multimodal AI APIs.The problem: I kept running into JSON responses stuffed with Base64-encoded images. Every time, I had to copy the string, find an online decoder, paste, preview. Repeat for each image. It got old fast.So I made ViewJSON. Paste your JSON, and it automatically detects and renders Base64 media inline: images, audio, video, even PDFs. No more copy-paste decode loops.Other things it does:
- Format&#x2F;minify JSON
- Paste an image, get Base64 string (reverse direction)
- Build API requests, copy as cURL
- Variable substitution for testing
- Share JSON via URLNo login, completely free. Would love feedback on edge cases I might have missed!

Show HN: I built a JSON viewer that decodes Base64 media inline

Hi HN — I built a small CLI called SkillLens to help answer: “What agent skills do I have installed, and are any of them sketchy?”A lot of agent ecosystems (Claude&#x2F;Codex&#x2F;OpenCode, etc.) store skills as folders with a SKILL.md. These files can contain surprisingly powerful instructions (and sometimes unsafe patterns), but they’re easy to forget once installed. We&#x27;re also tend to run them with --dangerously-skip-permissions and let them install whatever they want, but I got a bit anxious about it so decided to build a tool to have some peace of mind.I decided to not go with AST static check but instead use whatever CLI you have locally to validate it.SkillLens does two things:1. Discovery: it scans common local skill locations (configurable) and lists what it finds.
2. Optional audit: if you have an auditor CLI installed (claude or codex), it sends each SKILL.md (currently truncated to ~12k chars) to the auditor and asks for structured JSON output:- verdict: safe | suspicious | unsafe
- risk: 0–10
- summary + issues with evidenceIt also caches audit results locally so reruns won&#x27;t check skills again unless those were updated, you installed anything new or you explicitly asked it to do so with --force flag.Install&#x2F;run:npx skilllens scan
# or
pnpm dlx skilllens scanNotes &#x2F; caveats:- v0.1; I’m still iterating on the prompt&#x2F;schema and the “what counts as suspicious” heuristics.
- Today it sends the skill text to whatever your auditor CLI uses (so treat it like sharing the skill contents with that provider). “Redacted evidence extraction” is planned, but not implemented yet.
- If the auditor CLI isn’t installed, it still produces a scan report and marks audits as skipped.

Show HN: SkillLens – scan and audit locally installed agent skills

Creator here. I built ChartGPU because I kept hitting the same wall: charting libraries that claim to be &quot;fast&quot; but choke past 100K data points.The core insight: Canvas2D is fundamentally CPU-bound. Even WebGL chart libraries still do most computation on the CPU. So I moved everything to the GPU via WebGPU:- LTTB downsampling runs as a compute shader
- Hit-testing for tooltips&#x2F;hover is GPU-accelerated
- Rendering uses instanced draws (one draw call per series)The result: 1M points at 60fps with smooth zoom&#x2F;pan.Live demo: <a href="https:&#x2F;&#x2F;chartgpu.github.io&#x2F;ChartGPU&#x2F;examples&#x2F;million-points&#x2F;" rel="nofollow">https:&#x2F;&#x2F;chartgpu.github.io&#x2F;ChartGPU&#x2F;examples&#x2F;million-points&#x2F;</a>Currently supports line, area, bar, scatter, pie, and candlestick charts. MIT licensed, available on npm: `npm install chartgpu`Happy to answer questions about WebGPU internals or architecture decisions.

Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)

Hi HN. I built an API to solve personality drift in long LLM contexts.Unlike RAG (which handles facts), AIIM enforces behavioral state using 12 parametric locks (Logic, Empathy, etc.) and tracks emotional decay. It forces the model to stay &quot;in character&quot; even when the context window fills up.Free API for devs: <a href="https:&#x2F;&#x2F;ai-im.tech" rel="nofollow">https:&#x2F;&#x2F;ai-im.tech</a>

Show HN: AIIM – Parametric Identity Engine for Consistent NPCs

I built this after realizing that most reading tools
optimize for extracting information, not for reading itself.This app is designed for staying with a book:
- minimal, distraction-free UI
- offline-first, local data
- no forced notes, no summaries, no gamificationYou open a book, read a chapter, and think.
Nothing interrupts you unless you ask.It’s not about reading faster.
It’s about reading more attentively.I’m sharing this to get feedback from people
who care about books and deep reading.

Show HN: A quiet, offline-first reading app for reading books

Hey HN — I built Take a Break, a Chrome extension that starts a timer on streaming sites and gently reminds you when it’s time to sleep. It shows a countdown with snooze, and a midnight reminder without auto‑close. You can customize timers and sites.Would love feedback on UX and if the reminders feel too aggressive or too soft.Demo &#x2F; listing: <a href="https:&#x2F;&#x2F;chromewebstore.google.com&#x2F;detail&#x2F;ehhhemnkemficpheoidniabkoohmolaj" rel="nofollow">https:&#x2F;&#x2F;chromewebstore.google.com&#x2F;detail&#x2F;ehhhemnkemficpheoid...</a>Repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;hardiksondagar&#x2F;take-a-break" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;hardiksondagar&#x2F;take-a-break</a>

Show HN: Take a Break – a gentle extension to stop autoplay late at night

Show HN: Rails UI

Show HN: RatatuiRuby wraps Rust Ratatui as a RubyGem – TUIs with the joy of Ruby

Show HN: Differentiable Quantum Chemistry

I saw a tweet that tinygrad is so good that you could make a graphics library that wraps tg. So I’ve been hacking on a gtinygrad, and honestly it convinced me it could be used for legit research.The JIT + tensor model ends up being a really nice way to express light transport all in simple python, so I reimplemented some new research papers from SIGGRAPH like REstir PG and SZ and it just works. instead of complicated cpp its just a 200 LOC of python.

Show HN: High speed graphics rendering research with tinygrad/tinyJIT

I am developing an architecture called Sparse Gated Resonance (SGR). It is a sequence modeling approach designed to avoid the quadratic scaling of traditional Self-Attention. I have been benchmarking a 722k-parameter SGR against a 921k-parameter Transformer on Victor Hugo’s &quot;Notre-Dame de Paris&quot; (English).The SGR replaces the attention mechanism with a &quot;Causal Pulse.&quot; It uses gated 1D convolutions to generate a navigation vector that resonates against a brain-map of character embeddings. This allows the model to maintain a &quot;Living Cell&quot; state that updates with linear complexity.Full source and implementation: <a href="https:&#x2F;&#x2F;github.com&#x2F;MrPan2048&#x2F;GeometricTransformer&#x2F;" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;MrPan2048&#x2F;GeometricTransformer&#x2F;</a>Benchmarking Data (Notre-Dame de Paris):STEP 3900 ARCH | LOSS | PPL | ENT | TIME
SGR | 1.4481 | 4.26 | 1.5476 | 19.0ms 
STD | 2.0275 | 7.59 | 2.1476 | 40.3msSemantic Comparison (Generation from &quot;Quasimodo&quot;):SGR: &quot;Quasimodo. Then minds that the accasteady which which the&quot; 
STD: &quot;Quasimododo ng, o uer tre the todo hemo’He wand at tine.&quot;Technical Observations:Computational Efficiency: SGR maintains a significant latency advantage, consistently running at ~19ms compared to the Transformer&#x27;s ~40ms. This confirms the efficiency of the linear pulse over quadratic attention.Convergence Quality: By Step 3700, SGR reached a Perplexity (PPL) of 4.46, whereas the Transformer lagged at 8.36. SGR successfully produces recognizable English phrases and punctuation, while the Transformer still exhibits &quot;stuttering&quot; artifacts (e.g., &quot;Quasimodododod&quot;).Entropy Stability: SGR has stabilized at an entropy of ~1.54, which represents the optimal &quot;Mastery Zone&quot; for English text. The Transformer’s higher entropy (~2.14) correlates with its lack of structural coherence.I am seeking an endorsement to publish a formal paper on this architecture to arXiv (CS.ML). I believe these results demonstrate that &quot;Living Cell&quot; resonance models can outperform Attention in parameter-constrained and latency-sensitive environments. If you are a researcher willing to endorse or review the mathematical formalization, please contact me via GitHub.

Show HN: SGR – A Linear-Complexity "Living Cell" Outperforming Transformers

I made this for myself, and it seemed like it might be useful to others. I&#x27;d love some feedback, both on the threat model and the tool itself. I hope you find it useful!Backstory: I&#x27;ve been using many agents in parallel as I work on a somewhat ambitious financial analysis tool. I was juggling agents working on epics for the linear solver, the persistence layer, the front-end, and planning for the second-generation solver. I was losing my mind playing whack-a-mole with the permission prompts. YOLO mode felt so tempting. And yet.Then it occurred to me: what if YOLO mode isn&#x27;t so bad? Decision fatigue is a thing. If I could cap the blast radius of a confused agent, maybe I could just review once. Wouldn&#x27;t that be safer?So that day, while my kids were taking a nap, I decided to see if I could put YOLO-mode Claude inside a sandbox that blocks exfiltration and regulates git access. The result is yolo-cage.Also: the AI wrote its own containment system from inside the system&#x27;s own prototype. Which is either very aligned or very meta, depending on how you look at it.

Show HN: yolo-cage – AI coding agents that can't exfiltrate secrets

I built this as a small side project to learn and experiment, and I ended up with this!I used a subdomain from my personal portfolio, and everything else runs on free tiers.The project uses Nuxt, SVG, Cloudflare Workers, D1 (SQL), KV, Terraform, and some agentic coding with OpenAI Codex and Claude Code.What started as a joke among friends turned into a fun excuse to build something end to end, from zero to production, and to explore a few things 
I’d never touched before.I’d really appreciate any feedback or suggestions.

Show HN: Open-source certificate from GitHub activity

Hey HN! I built Retain as the evolution of claude-reflect (github.com&#x2F;BayramAnnakov&#x2F;claude-reflect).The original problem: I use Claude Code&#x2F;Codex daily for coding, plus claude.ai and ChatGPT occasionally. Every conversation contains decisions, corrections, and patterns I forget existed weeks later. I kept re-explaining the same preferences.claude-reflect was a CLI tool that extracted learnings from Claude Code sessions. Retain takes this further with a native macOS app that:- Aggregates conversations from Claude Code, claude.ai, ChatGPT, and Codex CLI
- Instant full-text search across thousands of conversations (SQLite + FTS5)It&#x27;s local-first - all data stays in a local SQLite database. No servers, no telemetry. Web sync uses your browser cookies to fetch conversations directly.

Show HN: Retain – A unified knowledge base for all your AI coding conversations

Hi HN, we&#x27;re Sam, Shane, and Abhi.Almost a year ago, we first shared Mastra here (<a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43103073">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43103073</a>). It’s kind of fun looking back since we were only a few months into building at the time. The HN community gave a lot of enthusiasm and some helpful feedback.Today, we released Mastra 1.0 in stable, so we wanted to come back and talk about what’s changed.If you’re new to Mastra, it&#x27;s an open-source TypeScript agent framework that also lets you create multi-agent workflows, run evals, inspect in a local studio, and emit observability.Since our last post, Mastra has grown to over 300k weekly npm downloads and 19.4k GitHub stars. It’s now Apache 2.0 licensed and runs in prod at companies like Replit, PayPal, and Sanity.Agent development is changing quickly, so we’ve added a lot since February:- Native model routing: You can access 600+ models from 40+ providers by specifying a model string (e.g., `openai&#x2F;gpt-5.2-codex`) with TS autocomplete and fallbacks.- Guardrails: Low-latency input and output processors for prompt injection detection, PII redaction, and content moderation. The tricky thing here was the low-latency part.- Scorers: An async eval primitive for grading agent outputs. Users were asking how they should do evals. We wanted to make it easy to attach to Mastra agents, runnable in Mastra studio, and save results in Mastra storage.- Plus a few other features like AI tracing (per-call costing for Langfuse, Braintrust, etc), memory processors, a `.network()` method that turns any agent into a routing agent, and server adapters to integrate Mastra within an existing Express&#x2F;Hono server.(That last one took a bit of time, we went down the ESM&#x2F;CJS bundling rabbithole, ran into lots of monorepo issues, and ultimately opted for a more explicit approach.)Anyway, we&#x27;d love for you to try Mastra out and let us know what you think. You can get started with `npm create mastra@latest`.We&#x27;ll be around and happy to answer any questions!

Show HN: Mastra 1.0, open-source JavaScript agent framework from the Gatsby devs

Hi HN,I built this because I often work in coworking spaces or do screen sharing, and I&#x27;ve always had this fear of accidentally flashing my .env file with production secrets to the whole room (or recording).It’s a simple VS Code extension that opens .env files in a custom grid editor. It automatically masks any value longer than 6 characters so I can safely open the file to check keys without exposing the actual secrets.It runs 100% locally with zero dependencies (I know how sensitive these files are). It just reads the file, renders the grid, and saves it back as standard text.It&#x27;s open source (MIT) and I&#x27;d love any feedback on the masking logic or other features that would make it safer to use.Marketplace: <a href="https:&#x2F;&#x2F;marketplace.visualstudio.com&#x2F;items?itemName=xinbenlv.dotenv-mask-editor" rel="nofollow">https:&#x2F;&#x2F;marketplace.visualstudio.com&#x2F;items?itemName=xinbenlv...</a> Github <a href="https:&#x2F;&#x2F;github.com&#x2F;xinbenlv&#x2F;dotenv-mask-editor" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;xinbenlv&#x2F;dotenv-mask-editor</a>

Show HN: Dotenv Mask Editor: No more embarrassing screen leaks of your .env

Hey folks, I’m Hassan, one of the co-founders of Infracost (<a href="https:&#x2F;&#x2F;www.infracost.io">https:&#x2F;&#x2F;www.infracost.io</a>). Infracost helps engineers see and reduce the cloud cost of each infrastructure change before they merge their code.
The way Infracost works is we gather pricing data from Amazon Web Services, Microsoft Azure and Google Cloud. What we call a ‘Pricing Service’, which now holds around 9 million live price points (!!). Then we map these prices to infrastructure code. Once the mapping is done, it enables us to show the cost impact of a code change before it is merged, directly in GitHub, GitLab etc. Kind of like a checkout-screen for cloud infrastructure.We’ve been building since 2020 (we were part of YC W21 batch), and iterating on the product, building out a team etc. However, back in 2020 one of our users asked if we can also show the carbon impact alongside costs.It has been itching my brain since then. The biggest challenge has always been the carbon data. The mapping of carbon data to infrastructure is time consuming, but it is possible since we’ve done it with cloud costs. But we need the raw carbon data first. The discussions that have happened in the last few years finally led me to a company called Greenpixie in the UK. A few of our existing customers were using them already, so I immediately connected with the founder, John.Greenpixie said they have the data (AHA!!) And their data is verified (ISO-14064 &amp; aligned with the Greenhouse Gas Protocol). As soon as I talked to a few of their customers, I asked my team to see if we can actually finally do this, and build it.My thinking is this: some engineers will care, and some will not (or maybe some will love it and some will hate it!). For those who care, cost and carbon are actually linked; meaning if you reduce the carbon, you usually reduce the cost of the cloud too. It can act as another motivation factor.And now, it is here, and I’d love your feedback. Try it out by going to <a href="https:&#x2F;&#x2F;dashboard.infracost.io&#x2F;">https:&#x2F;&#x2F;dashboard.infracost.io&#x2F;</a>, create an account, set up with the GitHub app or GitLab app, and send a pull request with Terraform changes (you can use our example terraform file). It will then show you the cost impact alongside the carbon impact, and how you can optimize it.I’d especially love to hear your feedback on if you think carbon is a big driver for engineers within your teams, or if carbon is a big driver for your company (i.e. is there anything top-down about carbon).AMA - I’ll be monitoring the thread :)Thanks

Show HN: See the carbon impact of your cloud as you code

Show HN: Diesel-guard v0.5.0 – Lint Diesel/SQLx Postgres migrations (24 checks)

I been building a view into companies based on their job postings. I am hoping it is useful to get a quick snapshot on the company say for interview prep etc. Please be advised that it is still a WIP in terms of data quality. Example urls: <a href="https:&#x2F;&#x2F;jobswithgpt.com&#x2F;company-profiles&#x2F;openai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;jobswithgpt.com&#x2F;company-profiles&#x2F;openai&#x2F;</a> <a href="https:&#x2F;&#x2F;jobswithgpt.com&#x2F;company-profiles&#x2F;advanced-micro-devices-inc&#x2F;" rel="nofollow">https:&#x2F;&#x2F;jobswithgpt.com&#x2F;company-profiles&#x2F;advanced-micro-devi...</a>

Show HN: Company hiring trends and insights from job postings

Show HN: Yashiki – A tiling window manager for macOS in Rust, inspired by River

Hi HN! I built Ghibli Search, a semantic search engine for Studio Ghibli movie scenes (e.g. Spirited Away, My Neighbor Totoro, Howl&#x27;s Moving Castle, etc.).Describe a dreamscape like &quot;flying through clouds at sunset&quot; or upload an image, and it finds visually similar scenes from the films.Live demo: <a href="https:&#x2F;&#x2F;ghibli-search.anini.workers.dev&#x2F;" rel="nofollow">https:&#x2F;&#x2F;ghibli-search.anini.workers.dev&#x2F;</a>Full Cloudflare stack: Workers, AI Search, R2, Workers AIOpen source: <a href="https:&#x2F;&#x2F;github.com&#x2F;aninibread&#x2F;ghibli-search" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;aninibread&#x2F;ghibli-search</a>Would love feedback on the search quality and any ideas for improvements!

Show HN: Semantic search engine for Studio Ghibli movie

Show HN: Agent Skills Leaderboard

Show HN: Hyve – Parallel isolated workspaces for coding agents, multi-repo dev

Show HN: Skill & MCP server for searching and retrieving 200k+ icons

Hi Hacker News! I’m David Huie. I’m launching SpeechOS, a drop-in voice input SDK for web apps.I was inspired by Wispr Flow and wanted the same workflow inside business apps (CRMs, docs, forms, support tools), not just a standalone tool. It’s saved me a massive amount of time vs typing.How it works: add a couple lines of JS plus an API key, and SpeechOS shows a small mic widget on every text field.Live demo: <a href="https:&#x2F;&#x2F;www.speechos.ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.speechos.ai&#x2F;</a>(Click a text box and the mic widget appears; click the gear icon to see settings, custom vocabulary, and snippet configuration.)Users can:- Dictate: speak naturally, real-time voice to polished text (punctuation, no filler&#x2F;typos)- Edit: say “make it shorter”, “fix grammar”, “translate...”- Command: define Siri-style app actions (for example, “submit form”, “mark complete”), and we match intent to your commandsIt also supports:- Custom vocabulary: domain terms and names (product names, acronyms, jargon) so they transcribe correctly- Text snippets: saved reusable blocks of text you can insert by voice (for example, “my signature”, “disclaimer”, “my address”)Why: text entry speed and accuracy still matter for productivity tools. A large-scale text entry dataset with 37,370 participants showed an average typing speed of 36.2 WPM and ~2.3% uncorrected errors. In speech research, speech recognition was about 3× faster than keyboard input and had ~20.4% lower error rate for English text entry. (<a href="https:&#x2F;&#x2F;hci.stanford.edu&#x2F;research&#x2F;speech&#x2F;" rel="nofollow">https:&#x2F;&#x2F;hci.stanford.edu&#x2F;research&#x2F;speech&#x2F;</a>)SpeechOS is currently in beta and free for now. Sign up at <a href="https:&#x2F;&#x2F;app.speechos.ai&#x2F;accounts&#x2F;signup&#x2F;" rel="nofollow">https:&#x2F;&#x2F;app.speechos.ai&#x2F;accounts&#x2F;signup&#x2F;</a> and enter this HN-only beta code: HN-JFc74cVC (please don’t share outside HN).Links:SDK repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;speechos-org&#x2F;speechos-client" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;speechos-org&#x2F;speechos-client</a>Demo: <a href="https:&#x2F;&#x2F;www.speechos.ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.speechos.ai&#x2F;</a>Signup (code: HN-JFc74cVC): <a href="https:&#x2F;&#x2F;app.speechos.ai&#x2F;accounts&#x2F;signup&#x2F;" rel="nofollow">https:&#x2F;&#x2F;app.speechos.ai&#x2F;accounts&#x2F;signup&#x2F;</a>I’d love feedback on:1) Where this is most valuable in your stack (notes? docs? CRM entry? support macros?)2) What you’d want from voice commands&#x2F;snippets configuration3) What would make you comfortable shipping this (privacy&#x2F;security, latency, pricing)If you’re building anything in voice AI&#x2F;dictation, I’d be happy to swap notes (david@speechos.ai).

Show HN: SpeechOS – Wispr Flow-inspired voice input for any web app

Hi HN, I’m experimenting with a small Python library called PicoFlow for building
LLM agent workflows using a lightweight DSL.I’ve been using tools like LangChain and CrewAI, and wanted to explore a simpler,
more function-oriented way to compose agent logic, closer to normal Python control
flow and async functions.PicoFlow focuses on:
- composing async functions with operators
- minimal core and few concepts to learn
- explicit data flow through a shared context
- easy embedding into existing servicesA typical flow looks like:<pre><code> flow = plan &gt;&gt; retrieve &gt;&gt; answer
 await flow(ctx)
</code></pre>
Patterns like looping and fork&#x2F;merge are also expressed as operators rather than
separate graph or config layers.This is still early and very much a learning project. I’d really appreciate any
feedback on the DSL design, missing primitives, or whether this style feels useful
for real agent workloads.Repo: https:&#x2F;&#x2F;github.com&#x2F;the-picoflow&#x2F;picoflow

Show HN: PicoFlow – a tiny DSL-style Python library for LLM agent workflows

Hey HN! I built TopicRadar to solve a problem I had with staying on top of what&#x27;s trending in AI&#x2F;ML without checking 7+ sites daily.<a href="https:&#x2F;&#x2F;apify.com&#x2F;mick-johnson&#x2F;topic-radar" rel="nofollow">https:&#x2F;&#x2F;apify.com&#x2F;mick-johnson&#x2F;topic-radar</a>What it does:
- Aggregates from HackerNews, GitHub, arXiv, StackOverflow, Lobste.rs, Papers with Code, and Semantic Scholar
- One-click presets: &quot;Trending: AI &amp; ML&quot;, &quot;Trending: Startups&quot;, &quot;Trending: Developer Tools&quot;
- Or track custom topics (e.g., &quot;rust async&quot;, &quot;transformer models&quot;)
- Gets 150-175 results in under 5 minutesBuilt for the Apify $1M Challenge. It&#x27;s free to try – just hit &quot;Try for free&quot; and use the default &quot;AI &amp; ML&quot; preset.Would love feedback on what sources to add next or features you&#x27;d find useful!

Show