Dimly AI — AI Music Video Maker & Generator
Turn a Suno track or your own song into a directed music video. The crew locks a style, directs each section, and cuts the whole song to your original audio.
DEV TRENDS / DAILY DISCOVERIES
Daily discoveries from developer communities, independent makers, and open source.
Turn a Suno track or your own song into a directed music video. The crew locks a style, directs each section, and cuts the whole song to your original audio.
PromptImage is a free AI image and video studio that turns text prompts or reference photos into polished visuals. Browse a prompt library of real creative examples, then generate with models including Nano Banana, Flux, Seedream, and Veo.
Story Claw is an open-source AI novel-to-video tool for short dramas, MG animation, and storyboards. It runs on Windows and connects to self-hosted ComfyUI/LTX rendering; storyclaw-templates add complex MG animations via function calls.
Mimi is a free, open-source real-time subtitle translator that turns foreign-language video or livestream audio on your computer into floating subtitles. Supports Apple silicon macOS 13+ and Windows x64; you supply cloud API credentials.
Give it a topic and it writes the script, finds visuals, adds voiceover, burns in subtitles, and outputs a finished short video plus publishing copy, with a zero-cost default path for the first video.
Learn JEV API integration with typed decisions, practical examples, and safe fallbacks. Explore Noul, Choice, and Score, then start building on BeatAPI.
One command audits AI projects for compliance red flags—cross-border data, hardcoded keys, and personal information exposure—maps findings to China's Cybersecurity Law, PIPL, and MLPS 2.0, and scores them. Zero-dependency, open source.
Attunz is a free 38MB no-signup macOS voice input app for Vibe Coding. It uses screenshot-free context awareness and a memory/dream system to transcribe terms and names. Supports offline local models or cloud APIs, incl. Codex/Grok subs.
macOS keyboard navigation app: click, drag, scroll and target UI elements without a mouse via Accessibility API, on-device Apple Vision OCR when accessibility data is missing, and Grid mode. Vim-style scrolling; open source; macOS 14+.
Free open-source Windows 11 desktop icon manager as a Fences alternative: group shortcuts/files by rules without moving them, tab groups, map real folders, hide on double-click, save/restore layouts. Rust, Apache-2.0.
Fan-made Delta Force browser mini-game hub. Havoc Coin War, a PVE tactical auto-battler, has you recruit operators, set rows, and build synergies; other games include stacking elimination, a loot-run simulator and reflex tests.
Open-source MIT DeepSeek-powered local coding assistant. Self-built agentic engine: streaming loop, file/command/search/web tools, approval gateway, MCP/Hooks/Skills/sub-agents. One engine drives CLI, VS Code plugin, and HTTP API.
Browser tool using your camera to check 8 real-time posture metrics (distance, head tilt, forward head, slouching, shrugging, leaning, blinking) with sitting and 20-20-20 eye-rest reminders, floating window, calibration, bilingual, dark mo…
Enterprise Java SDK that wraps backend APIs into secure native CLI tools and Skill packages for AI Agents and engineers: progressive schema discovery, @AgentCommand hints, risk declarations, dry-run previews, write confirmation guards, mas…
Figma plugin: say "add a circle here" to draw it. Jev resolves semantics via parallel multiple-choice questions (action, shape, position); code validates and calls the Figma API. Low confidence re-asks; unclear skips execution.
Peekd: lightweight single-binary web file server for fast local file browsing/sharing. Previews Markdown, HTML, XML, code, text, CSV, TSV, images, audio, video, PDF, ZIP/TAR/TAR.GZ. HTTP ranges, sendfile, auto dark mode.
Browser-based local video subtitling: Whisper runs via WebAssembly/WebGPU to transcribe, translate, style and burn subtitles into MP4 — no upload, signup or watermark. Also browser extension, desktop and CLI; exports SRT/VTT/ASS.
Private Agent Skills registry on your Cloudflare account. Manage skills on the web and install them into Claude Code, Cursor or Codex with one npx skills add. Speaks the skills.sh discovery protocol;
Reads Claude Code's local transcripts to warn at 90% quota usage, prompt a HANDOFF.md before limits hit, log the reset time, and auto-resume the same session with --resume afterwards.
Free macOS terminal/SSH client. Log in once to keep sessions alive across tabs; supports JumpServer with 2FA, jump hosts, SFTP, and live stats. Claude Code/Codex can reuse connected servers via a local interface without exposing passwords.
Local-first AI image workbench with Git-like version trees for prompt iterations and edits. Uses cheap API relays or local models, keeps assets local. Batch generation, inpainting, image text editing, outpainting, watermark removal.
AnyFileKit: 26 browser tools for compressing and converting images, PDF, video and Excel — PDF to JPG/Word, HEIC to JPG, video to MP3/GIF. Files never leave the browser; no sign-up or watermarks. AI translations send extracted text only.
Gear is a framework for evaluation-driven AI agent optimization. It runs tasks, collects traces/scores, lets an optimizer agent revise the harness: prompts, tools, skills, workflows, context, then re-evaluates and keeps better versions.
ToolMelo is a free, no-registration, ad-free online file tool site that processes PDFs and images locally in the browser.
AI job-search workbench keeping resumes and job data in one local DB file. It separates facts from wording, cross-checking AI-written resumes against your records to flag fabricated claims, plus collection, tracking and mock interviews.
Canary: zero-dependency Python framework where each service has init/start/stop and dep()-declared typed deps. Builds a dependency graph, starts in topological order with independent services concurrent, reclaims resources on failure.
Local-first job workbench for Chinese job seekers, aggregating BOSS Zhipin, Zhaopin, Liepin, and 51job; filters risky roles, scores commute/salary/match/employer/region, compares two jobs, edits/syncs resumes, and shows message badges.
nim_duilib is a C++ cross-platform UI library based on duilib. It supports Windows, Linux, macOS and FreeBSD, uses DirectUI/XML layouts, rich controls, Skia rendering, i18n, skins and DPI, simplifying desktop UI development.
Client SDK by Trade Data Pte. Ltd. (Singapore) for global customs bill of lading and trade data across 200+ countries: importer/exporter profiles, HS codes, logistics flows via standardized APIs and regional nodes. Python and Node.js.
Cloudflare Workers (TypeScript) reimplementation of SyncClipboard's server. Official clients (v3.1.1+) work unchanged with clipboard sync, WebDAV endpoints and a Web history manager. No VPS; fits Cloudflare's free tier.
Self-hosted AI coding agent: label a GitHub Issue ai-ready and it writes code, opens a PR, reviews the diff against the issue's acceptance criteria separately, merges only on pass, then tags and releases. Uses your own API key or Codex.
Agent Skill for Claude Code, Codex, Cursor: given a photo, it returns camera coordinates, heading, error radius and a satellite evidence image. The model only reads images; 20 Python scripts search OSM, elevation and street-view data.
Local-first campus AI agent for students: one-sentence queries for timetables, grades, exams, notices and todos via terminal, web or QQ bot. Credentials AES-256-GCM encrypted on-device, no telemetry; 31 tools; pluggable school adapters.
Learn it. Build it. Ship it for others.
Hindsight: Agent Memory That Learns
A unified library of SOTA model optimization techniques like quantization, distillation, pruning, neural architecture search, speculative decoding, etc.
Fix X/Twitter and Bluesky embeds! Use multiple images, videos, polls, translations and more on Discord, Telegram and others
No project description yet.
An offline-first photo manager for large local libraries
Diffusion model(SD,Flux,Wan,Qwen Image,Z-Image,...) inference in pure C/C++
AI Agent Assistant & development framework that integrates lots of IM platforms, LLMs, plugins and AI feature, and can be your openclaw alternative. ✨
A searchable Linux command encyclopedia with manuals, detailed explanations, and learning resources for quickly looking up and understanding Linux commands.
A Chinese-localized Clash for Windows project, now providing an improved version that replaces the core with Mihomo and adds compatibility with modern protocols such as AnyTLS (still in testing).
A quantitative research project that reproduces sell-side financial engineering research reports.
Open-source, free, cross-platform Bilibili video downloader with multithreaded acceleration, separate audio/video streams, danmaku and metadata retrieval, custom naming and categorization.
Free, open-source offline OCR software. Recognizes text from screenshots, batch images, and PDFs; can exclude watermarks/headers/footers; scans or generates QR codes; includes multilingual recognition libraries.
RuoYi-Vue Pro: Spring Boot, MyBatis Plus, Vue/Element admin framework + WeChat mini program. Supports RBAC/data permissions, SaaS multi-tenancy, Flowable workflows, third-party login, payment, SMS, mall, CRM, ERP, MES, IM, AI models, IoT.
Tests Cloudflare CDN latency and speed to find the fastest IP; also supports other CDNs and sites with multiple resolved IPs.
Enterprise AI low-code platform that generates full-stack code or an entire system from one prompt. AI Skills build flows, forms, reports and dashboards;
A community-run repository where companies and teams in blockchain and other industries can share hiring information; all job posts are published across Rebase community media, free of charge.
minimi is personal intelligence that comes to you. We are building a world where before you've said a word, AI already understands you and frees you from the work of managing your own life, so you can spend that time actually living it.
Most businesses juggle a different tool for every channel — one for phone support, another for WhatsApp, another for the website chatbot.
Semrush and Ahrefs cost $130+/month for work that's mostly deterministic: crawling a site, checking meta tags, validating schema, checking what AI crawlers can see. jev-seo does that in a free Rust CLI.
AI coding agents made code fast; delivery still crawls, because coding is one seat in the product cycle. Autonomous Product Delivery runs the whole loop as one system on your real codebase.
AI agents are sprawling across clouds, SaaS tools, Kubernetes, and devices, with no common layer to govern them.
NotchPop turns your MacBook notch into a Dynamic Island: music, a file shelf, clipboard history, focus timer, calendar, weather, AI coding stats and revenue. Works on Macs without a notch too. Native SwiftUI, $3.99 once, 7-day free trial.
Coding agents push more code, so your CI runs more builds than ever. GitHub's runners are slow, so every extra build sits in the queue and your team waits with it.
Do you actually know who has access to your servers right now? CtrlOps gives you a security hardening score across 25 checks, a PDF audit report, and fix commands for every issue found.
Harness Manager is the App Store and control center for AI coding harnesses on Mac. Discover, install and update tools like Claude Code, Codex, OpenCode, Pi and more.
Maximem Synap is memory and context infrastructure for AI agents, so every conversation does not start from zero. It is the fastest and most accurate memory system on public benchmarks, 92% on LongMemEval and 93.
Two new things, both in beta. Financial plan: describe your money in plain words and Subscrr lays your year out month by month. Test a purchase or a month without income before you commit, then compare the plan with what actually happened.
The rebuilt Storytailor turns children’s ideas and drawings into recurring characters and illustrated stories for families, educators and care teams.
Every chore app asks how hard a chore is, once, for everyone. ChoreDivider asks each of you, in private: how draining is this for you? Ratings stay sealed until everyone is done.
Connect Floot to Claude or ChatGPT and build a full-stack app without leaving the chat. Floot gives your AI a ready-to-use workspace with a database, user logins, hosting, and a live URL - nothing to install or configure.
Your company has thousands of documents, but only one version of the truth. NOAN turns the facts your business has approved such as pricing, positioning, policies, products, customers, and more into a verified, versioned source of truth av…
A joke tool that abuses OpenType's ligature feature to mix fonts. It works pretty fast on client-side by loading Python in WASM.
I made a financial planning tool in a couple hours that can model various stages of life and potentially guide job searching and work timelines.
Own the stack: run Carbon on your own infrastructure — on-prem, in your VPC, or fully air-gapped. ERP, MRP, MES and QMS on Postgres you own, source available, with an AGPL-3.0 Community edition.
Hello HN! I've spent years debugging Windows crashes with tools that were either friendly but limited (e.g. Visual Studio) or powerful but archaic (e.g. WinDbg).
Hi HN, I just open sourced the DSL that our harness in grep.ai uses to turn repeatable parts of agent work into workflows.
Air-gapped file encryption packed into a single, self-decrypting HTML page. https://github.com/ApelegHQ/ts-cms-ep-sfx I was inspired by self-extracting archives.
Hacker News helps me stay in touch with the world of technology without putting in extensive web browsing work, and I love it for that reason.
Hi HN, this is Yarik and Vlad from VOYGR - we are building the tools for agents and apps to engage with local businesses. It all started with our own pain point at VOYGR: calling businesses to verify if they are open.
One endless vote: which tool would you rather delete from your life? Jira or Outlook? Teams or Slack? Unscientific, opinionated, and addictive.
I built an independent reimplementation of the QBasic 1.1 compiler, virtual machine, and IDE that runs entirely in the browser. TypeScript with a Svelte 5 frontend.
Most online language playgrounds stop at compiling code and printing text. But C3 contributor Manu Linares wanted C3’s playground to go further.
You start a run, tab away, and Claude Code sits waiting on one Allow. Crest puts the prompt on the notch: tap to answer, the reply lands in your terminal.
A quiet native macOS menu bar timer with a thin progress line.
Get the AI to guess the word. Type clues without using the word itself. Five words, 30 seconds each. Free to play, with a new daily game and unlimited practice.
A board, an open Norwegian mountain, and the sea below. Skate, wander, and discover at your own pace. Play free in your browser.
face-rating.net rates a frontal photo on a 10-point scale using an in-browser model; photos and intermediates never leave the device.
macOS native client for GitHub Projects that automates Issue status from linked PRs: Draft PR sets In Progress, review-ready sets In review, Done only after all PRs merge. Local-first caching, menu bar issue view, Backlog support.
Open-source gateway between OpenAI-compatible clients and local/cloud models: safety rules route requests to cloud, locally filtered context, or a local model, with fallback. Supports Ollama, tool calls, metrics; not yet production-ready.
25 年年底,我当时还在上班,有一天领导让我接手一个我不了解的项目,上来就要改接口,于是就有了用抓包工具从 app 端抓包看涉及的功能都调用了哪些接口,这才有了后续我们做抓包工具的灵感,后来我想用到一些功能但试了几款都没有,就更想自己做一款了。 25 年我在 app store 看到的抓包工具还不到十款吧(现在再去看就多了),基本长得都差不多,功能也差不多,体验也差不多。一看就知道都是基于开源的版本改的。 我感觉这是个机会吧,虽然是小众需求,但能选择的产品不多,而且现有产品也不好用,于是就想开发一款比现有产品都好用的抓包工具:UI 简洁、容易上手、操作简单。把 90%常用的功能做得足够简单,让用户使用不需要看说明书,并且操作步骤能一步完成就不要让用户操作两步。 第一个版本只做了几个核心功能,但用了一个月时间写完,因审核问题,又延迟一个月,到 26 年 1 月低才上架成功。 当时我其实没有了解到有 Proxyman 这款产品,我发布后不久才发现的,因为我没去做什么调研,只在 AppStore 上搜索有哪些抓包工具,而 Proxyman 又没有上架中国。 所以后面当我了解到 Proxyman 后,也是看 Proxyman 提供哪些功能我们还没有的就加上了,ApiCatcher 的功能才慢慢增多。 我并不想在价格上做优势,面向全球用户,Proxyman 定价多少我们也定价多少,我们想卷的是用户体验! 这半年我大部分时间都投入到了 ApiCatcher ,从开发 iOS 版本,到开发桌面端,再到 Android 端,目前已经支持 iOS 、Android 、Windows 、macOS 四个平台。也做了 Burp Suite 插件,能够在移动设备上实时捕获 HTTP/HTTPS 流量并实时传到 Burp Suite 上分析。还开放了实时同步协议,让所有用户都能实现自己开发接收端,实现类似 Burp Suite 插件这样的需求,这其实也是用户向我们提的需求。 在完成几个端的开发后,我们开始优化产品的一些细节体验,细节是最难描述的,但如果用过 ApiCatcher ,应该能感受到一些细节上的用心,举几个例子: - 编辑请求头时会提供常见请求头 key 和 value 选项(能选就不输入,因为记不住,还得去搜索复制)。 - 写脚本提供 AI 生成,APP 将请求自动生成 API 文档加入系统提示词,用户只需要描述需求就能生成脚本,生成的脚本还可以从历史请求中选择一个请求测试,不仅能查看结果还能看脚本打印的日志。 - 写正则表达式不会写? AI 一键生成,生成后可以选择一个请求测试,测试失败可以一键修复。 ..... 这样的细节非常多,在此不过多介绍,说再多不如上手体验。 Proxyman 做的一些 AI 功能我们并不想做,不想为了做而做。我们加入 AI 更想让 AI 做擅长的事情、做真正用户需要的事情:帮生成正则表达式、帮生成 cron 表达式、帮生成脚本等。 抓包工具那么多,但用户需求的功能就那几个,所以我们要做的并不是跟 Proxyman 等抓包工具比谁的功能更多、比谁的功能更强,而是在用户实际需求、常用的功能上,比谁的体验用户更喜欢。当然,我们不会做推广,知道 Proxyman 的用户并不知道 ApiCatcher 。 [ApiCatcher]( )的抓包功能是免费的,欢迎大家下载体验! 这里也赠送几个终身会员免费优惠链接:(在 iOS 上点击链接跳转到 AppStore 领取) 4XNLWHMLM3TY4RAJTT, 3AAEL7TL8EK67WNEM4, MAR8LAE8XMNLW4JYPE, 3WA83RPW6M4633XNAW, FPHFLNK8TEWT6T7X43, NERKHXYLENPJ4MFK4N,
Originalsn-motion-html is a Skill for building scroll-driven immersive web narrative videos. The Odyssey case covers asset prep, storyboards, character consistency, AI generation, FFmpeg processing, web assembly, and local preview.
An LLM-powered MUD framework with customizable content packs, a mobile-friendly web client, AI narrative styles, and NPC memory/tasks. It includes a xianxia pack with races, cultivation realms, items, skills, events, regions, and NPCs.
Converts PDFs, DOCX and HTML to Markdown for organizing into Obsidian or Notion. Text-layer PDFs run locally in the browser (up to 50 MB/500 pages, no upload); scanned PDFs and images can use optional cloud OCR.
EVASafe is a fully offline Android local password manager. Passwords are stored only on your phone. It is also notable for being written entirely by an agent.
PlainMote shares text, files, and configs via public URLs, avoiding WeCom long-text document conversion and social-platform content snooping. Use it without login; log in to control shares and add media/files. Quotas are small.
mcppls — a C++20/23 modules language server: one set of module semantics on GCC, Clang, MSVC and MinGW, for editors (VS Code, Zed, CLion, Neovim), AI Agents (MCP) and CI
Try another keyword or category.
15 repositories were already featured in the last 3 reports, so they are not repeated in today’s discoveries.
Google's open agentic orchestration runtime
The Office Harness for AI Agents — Spreadsheets, Docs, Slides, Canvas, Relational Tables, and PDF in one runtime.
An agentic skills framework & software development methodology that works.
DEV TRENDS / COMMUNITY
Discuss this edition
Sign in with GitHub to share a thought, question, or useful follow-up.