Tools Bench.

Product launches and open-source repos with enough signal to earn a second look.

Last Brew Time: Aug 21, 2026, 10:57 AM PT

Insight

Much of what shipped this run is not new AI capability so much as the same handful of frontier models repackaged behind a friendlier interface

01.MEETING-NOTETAKER WRAPPERS AROUND ZOOM/MEET/TEAMS

02.LORA REMIX DEMOS STACKED ON A COUPLE OF BASE VIDEO AND IMAGE MODELS

03.OBSERVABILITY LAYERS BOLTED ONTO AGENT TRAFFIC AFTER THE FACT

Featured

GitHub275.5K

obra/superpowers

An agentic skills framework & software development methodology that works.

youtube
Claude Code + SUPERPOWERS = The End of Vibe Coding? (Full Tutorial)

Market Signal

Why It Has Market Pull

Superpowers is a genuinely viral open-source skills framework and development methodology for Claude Code and other coding agents, with extraordinary GitHub growth (over 275,000 stars) since its creation in October 2025. Real, sustained builder momentum is visible across an active issue tracker, a growing plugin ecosystem, and native integrations shipped for competing agent CLIs, making this clearly worth a closer look.

  • Over 275,000 GitHub stars and 24,600 forks in under a year since the repo was created (October 2025)
  • Nearly 300 open issues and roughly 200 open pull requests, showing sustained day-to-day contributor and user engagement rather than a one-time spike
  • Native support has been added for Codex, OpenCode, Gemini CLI, Qwen Code, and GitHub Copilot CLI, evidence of real cross-tool demand beyond Claude Code alone
  • Spawned a companion ecosystem (superpowers-lab, superpowers-skills, superpowers-developing-for-claude-code) and is listed on multiple third-party skill marketplaces (Claude Bazaar, AugmentClaude)
  • Composable skills (brainstorming, TDD enforcement, git-worktrees) are cited by independent write-ups (Medium, Dev.to) as changing how people work with coding agents

feedbacks

What People Are Saying

  • "feat: Adds native Qwen Code CLI support following the same patterns as Codex and OpenCode installers"GitHub pull request

  • "feat: add GitHub Copilot CLI support"GitHub pull request

  • "feat: Adds native Antigravity support"GitHub pull request

  • "feat: Leverage Claude Code agent teams for parallel plan execution"GitHub pull request

  • "Production-Tested Improvements: Mechanical Enforcement & Safety Gates"GitHub pull request

  • "Superpowers for Claude Code"Medium article

  • "Claude Code Superpowers for Engineering-Grade Development"Dev tutorial site

GitHub228.9K

mattpocock/skills

Skills for Real Engineers. Straight from my .agents directory.

youtube
mattpocock/skills: A complete AI Coding workflow, end-to-end

Market Signal

Why It Has Market Pull

Matt Pocock's personal Claude Code / agent skills collection has become one of the most explosively adopted open-source repositories in the space, growing to 229,000+ stars and nearly 20,000 forks in about six and a half months, and it is available both as an official Claude Code marketplace plugin and as an editable script install, though independent commentary questions whether its elaborate skill phrasing is proven to outperform simpler instructions.

  • Over 229,000 GitHub stars and 19,600 forks as of August 2026, up from roughly 143,000 stars only a few months earlier
  • Repository created February 2026 and still receiving daily commits and new issues in August 2026, showing sustained not one-off momentum
  • 374 open GitHub issues, indicating an active user base filing bugs and feature requests rather than a dormant repo
  • Distributed both as an official Claude Code marketplace plugin (auto-updating) and as an editable script-copied version for customization
  • Hacker News commenters pushed back that no controlled comparison shows the elaborate skill phrasing actually outperforms plainer prompts

feedbacks

What People Are Saying

  • "This is why the grill me skill went viral"HN post

  • "there's no strong reason to trust an LLM's own account of what makes it perform better"HN comment

  • "phrasing like "walk down each branch of the decision tree" sounds more rigorous than plain versions, but nobody had run a controlled comparison"HN comment

  • "/to-tickets can leave PRD requirements orphaned"GitHub issue

  • "improve-codebase-architecture can silently break the UI it's refactoring"GitHub issue

  • "Skills for Real Engineers Need Governance, Not Fandom"Dev.to article

  • "Matt Pocock's Skills Repo Is a Better Pattern Than Vibe Coding"Dev.to article

GitHub38.2K

PostHog/posthog

🦔 PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.

youtube
What is PostHog? (Official Demo & Tutorial)

Market Signal

Why It Has Market Pull

PostHog is a mature, well-funded open-source product analytics platform with 38k+ GitHub stars that has repositioned itself around AI-driven 'self-driving products' -- using agents to triage errors, session replays, and experiments and even open pull requests automatically -- making it one of the more credible AI-observability plays already embedded in thousands of engineering teams' workflows.

  • 38,235 GitHub stars and 3,230 forks, with active releases shipping the same day this was reviewed (posthog-cli v0.14.0, desktop v0.60.372)
  • 5,153 open GitHub issues reflect a large, actively engaged user base rather than a quiet repo -- one closed issue alone drew 113 comments
  • Recently expanded from analytics into full AI observability with a five-stage 'self-driving product' pipeline (ingest, group, research, assess, execute) that can open pull requests automatically for regressions
  • Steerable from Slack, web, desktop, and MCP, positioning it as agent-native infrastructure rather than a bolt-on dashboard
  • A deep technical write-up ('I rewrote PostHog's SQL parser, 70x faster') drew 155 points and 72 comments on Hacker News, showing engineering-community respect beyond marketing

feedbacks

What People Are Saying

  • "Improved funnel visualisation"GitHub issue

  • "Bug: PostHog consuming excessive browser memory"GitHub issue

  • "Event retention // delete events after a certain timeframe"GitHub issue

  • "I rewrote PostHog's SQL parser, 70x faster, while barely looking at the code"HN post

  • "Posthog NPM packages are compromised"HN post

  • "Inside PostHog: SSRF, ClickHouse SQL Escape and Default Postgres Creds to RCE"HN post

  • "Posthog/.cursorrules"HN post

GitHub16.2K

RyanCodrai/turbovec

A vector index built on TurboQuant, written in Rust with Python bindings

youtube
GitHub - RyanCodrai/turbovec: A vector index built on TurboQuant, written in Rust with Python bin...

Market Signal

Why It Has Market Pull

TurboVec is a fast-growing Rust vector index (16,200+ GitHub stars in about five months) built on Google Research TurboQuant, backed by a 298-point Hacker News front-page thread, press coverage from MarkTechPost and Medium, and real integrations with LangChain, LlamaIndex, Haystack, and Agno — strong technical and community momentum for a solo-authored project, though it has no company or funding behind it and some commenters questioned the polish of its documentation.

  • 16,183 GitHub stars and 1,399 forks since a March 2026 creation date
  • Hacker News Show HN thread reached 298 points with active technical discussion
  • Claims 10M documents compressed from 31 GB to 4 GB versus FAISS, with published benchmark JSONs
  • Shipped a 1.0.0 release and active v5/v6/v7 index-format migration tooling
  • Integrates with LangChain, LlamaIndex, Haystack, and Agno, and was covered by MarkTechPost and Medium

feedbacks

What People Are Saying

  • "It would be nice to have the README be a little more human written for a project where you actually want people to adopt it"HN comment

  • "Wow! 4GB for 10 million documents...Can't wait"HN comment

  • "what could i use this for as part of my agentic workflow? codebase indexing? docs?"HN comment

  • "Who is this co-author called t <t@t>?"HN comment

  • "How this compares to Redis VSET?"GitHub issue

  • "Any plan to cut a release?"GitHub issue

  • "WASM support for Browser and Node.js?"GitHub issue

GitHub5.3K

Tencent/AI-Infra-Guard

A full-stack AI Red Teaming platform securing AI ecosystems via Agent Scan, Skills Scan, MCP scan, AI Infra scan and LLM jailbreak evaluation.

github
A.I.G Main Interface

Market Signal

Why It Has Market Pull

AI-Infra-Guard is a genuinely active, fast-growing open-source AI red-teaming toolkit from Tencent's Zhuque Lab, credited with helping NVIDIA, Google, and Microsoft patch AI infrastructure vulnerabilities, and it now ships one of the first public MCP supply-chain scanners alongside agent, skills, and jailbreak evaluation modules.

  • 5,307 GitHub stars and 509 forks, with one tracker reporting 435 new stars added in a single day -- genuine trending momentum, not a static mirror
  • Three releases shipped in about three weeks (v4.5.0, v4.5.1, v4.5.2), showing continuous engineering investment
  • Backed by Tencent's Zhuque Lab, which has a track record of helping NVIDIA, Google, Microsoft, and Hugging Face fix high-risk vulnerabilities
  • Ships one of the first publicly available MCP supply-chain scanners alongside Agent Scan, Skills Scan, and LLM jailbreak evaluation modules, addressing a genuinely underserved corner of AI security tooling

feedbacks

What People Are Saying

  • "Task creation failed: SSE connection timeout, please retry"GitHub issue

  • "MCP scan of remote service throws an error"GitHub issue

  • "Support for tamper-evident audit trails alongside security scanning"GitHub issue

  • "Feature request: add detection for OpenClaw"GitHub issue

  • "Scoring system is wrong, code needs correction"GitHub issue

  • "AI Infra Guard"HN post

  • "helped major vendors such as NVIDIA, Google, and Microsoft ... fix high-risk vulnerabilities"GitHub README

Sources

GitHub

harry0703/MoneyPrinterTurbo

113.7K

利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.

JuliusBrussee/caveman

100.0K

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

santifer/career-ops

67.3K

Open-source AI job search: scan job portals, evaluate listings with a structured A-F rubric into a 1.0-5.0 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)

makeplane/plane

56.9K

🔥🔥🔥 Open-source Jira, Linear, Monday, and ClickUp alternative. Plane is a modern project management platform to manage tasks, sprints, docs, and triage.

volcengine/OpenViking

31.5K

Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.

Product Hunt

HyNote for Mac

368

Solving the three biggest pain points in meeting software: privacy, friction, and cost. By running speech-to-text entirely on-device, your confidential discussions never leave your hardware, ensuring total data privacy. It operates invisibly without sending an awkward bot into your attendee list, capturing system audio directly across Zoom, Google Meet, and Microsoft Teams.

Grok 4.6

297

Continuous reasoning meets real-world execution. Grok 4.6 brings major upgrades to agentic workflows, software engineering, and interactive web application generation—at an unchanged, cost-efficient rate of $2 / $6 per 1M tokens. Build long-running AI agent workflows with Grok 4.6 on xAI API and partner networks!

Checksum AI

233

Checksum is an AI-native continuous testing platform for engineering teams shipping faster than manual QA can keep up. It generates, runs, and auto-heals end-to-end and API tests on every pull request, all as standard Playwright code in your own repo. When a test fails, Checksum tells you whether it found a real bug or a stale test, then fixes the false failures so your suite keeps pace with your coding agents.

The New Calendly

208

Meet Calendly Notetaker and Callie AI assistant: two new AI products built to handle the work before, during, and after meetings. Meetings are essential, but the work around them adds up fast. 😅 That's why we're launching an evolved Calendly. One that handles not just scheduling, but all of the work before, during, and after your meetings, so you can focus on what matters.

MeetStream AI

179

Agent-first infrastructure for meetings. One API to capture 50+ real-time data points from Zoom, Google Meet, and Teams, with built-in voice infrastructure so your agent joins as a participant, listens, speaks, and acts while the call is happening.

Prized

160

Prized lets ops, support, and finance teams build and ship secure internal tools with AI. Company data pre-connected and scoped, an audit trail on every access, one-click deploy behind your company sign-in.

YC Launch

The Last Accounting Company you’ll ever need

2

Agent-native accounting firm that replaces the legacy software and the outsourced service. Last Accounting Company · Summer 2026 · B2B Tags: Artificial Intelligence, Fintech, Finance. Website: https://lastaccountingcompany.com/

Hacker News

Proliferate- open-source, self-hostable Codex for any coding agent

11 pts

Hi HN- I'm Pablo, the founder of Proliferate! Proliferate ( https://github.com/proliferate-ai/proliferate ) is an open-source, self-hostable AI IDE that lets you work and automate tasks with Claude Code, Codex, OpenCode, Cursor, and Grok in one place. Here's a quick 2m demo of how we use Proliferate to build Proliferate: https://www.youtube.com/watch?v=tGNX0oaWmBY I started building Proliferate after my team onboarded to OpenAI Codex. Within days, we were using it for everything: using computer... (11 points, 5 comments).

MCP app for Android, drive apps via AI (no root, PII redacted locally)

6 pts

Hi HN, author here. Android Remote Control MCP is an MCP server that runs directly on your Android phone (no root, no ADB, no computer in the middle) and lets an AI agent drive real apps the way a human would: it reads the screen through the accessibility tree, taps, types, scrolls, optionally also screenshots. I spent a lot of time optimizing tool usage and token consumption, and making it work not only via local harnesses but also via Claude.ai / Claude Desktop and chatgpt.com (if you have the... (6 points, 1 comments).

A desktop fly drawn to the scent of vibecode

18 pts

It is a fork of https://github.com/DenisSergeevitch/desktop-fly , but with an important update. Now the fly can pick up the scent of the codebase with its neurons and fly straight to the source code of your B2B AI SaaS startup. It has learned to scan its surroundings for agent markers: AGENTS.md, CLAUDE.md, .cursor/rules, .kiro/steering, and forty others. Anything on the screen that points to these markers becomes a source of the scent - an editor window with an open project, a line in Finder, o... (18 points, 8 comments).

LilScript makes JavaScript libraries smaller

14 pts

https://yeargun.github.io/lilscript/ LilScript is a typed, compression-first language that compiles into js and sometimes into exec(will be more stable in future). The compiler mangles, reshapes the program into optimized js that happens to be 5-15% smaller (after gzip/br compression or raw) compared to the best performing JS toolchains like oxc/esbuild/terser/.. ## What has been proven to work with LilScript? - makes VSCode's core js modules 20% smaller on average - makes the world's most perfo... (14 points, 2 comments).

HarnessRouter: Unified interface for agent harnesses

10 pts

Hey HN! We are building HarnessRouter, a canonical API for running Codex, Claude Code, Hermes, and other managed agent harnesses as your product backend. Before building HarnessRouter, I used to build our own agent harness for our products. I tried LangGraph, agent SDKs from different vendors, pydantic, LLM tool use / function call, and so on. It's a very heavy lifting engineering effort, and I am disappointed about the agent deliveries compared to what Codex, CC can deliver. That changed my min... (10 points, 14 comments).

HF Spaces

Qwen-Image-Edit-2511-LoRAs-Fast

2.6K likes

Demo of the Collection of Qwen Image Edit LoRAs Qwen-Image-Edit-2511-LoRAs-Fast is a Hugging Face Space tagged with gradio, mcp-server, region:us. It has 2637 likes on Hugging Face.

Agent Memory Leaderboard

716 likes

Unified memory evaluation · Results expected August 12. Agent Memory Leaderboard · 记忆之巅 A unified, open, and reproducible evaluation platform for long-term memory systems and memory-enabled agents. Agent Memory Leaderboard (AML) compares research methods and commercial products under one evaluation contract. Candidate systems implement memory Add and Search; the official platform fixes Answer, Eval, datasets, models, configurations, result review, and publication. > First public release: The inaugural verified leaderboard is expected to be published on August 12, 2026. > 首期发布： 首期经核验榜单预计将于 2026 年 8 月 12 日发布。 Results are separated along two independent dimensions. Textual and coding tasks use...

MiniMax H3 Turbo LoRA

261 likes

Video generation with a synchronized soundtrack MiniMax-H3 — unquantized, split across two Spaces Joint video and soundtrack out of a single denoising pass, at bfloat16 with no quantization anywhere. This Space is the denoising half: the 61.73 GiB transformer and the two autoencoders. The 62.14 GiB Qwen3-VL conditioner runs in qwen3vl-conditioner, which this Space calls over the gradio API for every request. The weights are the public MiniMaxAI/MiniMax-H3 diffusers checkpoint. MiniMax-H3 is 195.9 GiB in bfloat16 and a ZeroGPU Space is evicted at 150 GB of storage. An unquantized single Space is therefore impossible, which is why quantized demos of it run NVFP4 or float8 weights. Cut the Mini...

Free AI Humanizer

239 likes

Free AI humanizer - natural AI text generation & rewriting. Official Lynote landing page for the free AI humanizer — AI text generation and rewriting: paste AI-generated text from ChatGPT, Claude, Gemini, or DeepSeek and get natural, human-like writing in seconds — no sign-up required. The free web tool is powered by Lynote's own open-source model, Lynote/humanize-text-model: two lightweight bilingual (English + Chinese) T5-small checkpoints, MIT licensed, fine-tuned on a reproducible rule-supervised corpus. Run it anywhere: View the model → · Try the full product → This Space is currently a static entry page: Hugging Face now hosts interactive Gradio demos on free CPU only with a PRO subscr...

MiniMax Music 3 Studio

232 likes

MiniMax Music 3 Studio — diffusers demo Streams full songs from lyrics + a structured caption using the MiniMaxMusic3Pipeline diffusers port. The input surface is a single Suno-inspired custom gr.HTML composer (Simple ↔ Studio modes, section-tag chips, structured-caption fields per the official prompting guide) that drives Gradio events via trigger()/props.value; styling uses only theme CSS vars so it follows the Citrus theme natively. Weights: MiniMaxAI/MiniMax-Music3 AoTI kernels: diffusers-internal-dev/MiniMax-Music3-aoti (compiled on RTX Pro 6000, matching ZeroGPU hardware) Generation streams chunk by chunk with a configurable playback headroom. The 8B language-model stage runs eager on....

OpenVuln

115 likes

Find bugs in your repository with GLM This Space is built automatically from the root Dockerfile and serves the Vite application with nginx on port 7860. Optional Space build variables: VITEAPIBASEURL — API origin; defaults to https://openvuln.vulnhunter.pro. VITEGITHUBREPOURL — source repository linked from the interface. OpenVuln is a Hugging Face Space tagged with docker, region:us. It has 115 likes on Hugging Face.