Tools Bench.

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

Last Brew Time: Oct 9, 2026, 11:23 AM PT

Insight

Across this run, a quieter trend than the obvious agent hype is builders relabeling someone else's model as their own launch

01.SAME-DAY QWEN-IMAGE-2.1 REPACKAGING

02.AGENT SKILLS AS THE NEW PACKAGING FORMAT

03.BENCHMARKS BUILT BEFORE THE PRODUCT THEY MEASURE EVEN HAS ONE

Featured

GitHub103.8K

addyosmani/agent-skills

Production-grade engineering skills for AI coding agents.

x
Addy Osmani shows how he uses his agent-skills repo for a 24/7 multi-agent system

Market Signal

Why It Has Market Pull

A genuinely viral open-source skills pack from a credible, well-known engineer (Addy Osmani, ex-Google Chrome), with explosive star growth and a substantive, two-sided Hacker News debate — real signal, not just name recognition.

  • 103,913 GitHub stars and 10,863 forks, for a repo created in February 2026 — roughly 8 months old
  • 24 skills covering the full engineering lifecycle, compatible with Claude Code, Codex, Cursor, and claimed 70+ agents
  • 138 open GitHub issues currently tracked, including active community proposals
  • Covered independently across at least 7 blogs/newsletters within months of release
  • Author is a recognized, verifiable industry figure (Google Chrome engineering/DevRel, published author), adding credibility beyond typical anonymous OSS projects

feedbacks

What People Are Saying

  • "LLMs aren't perfect rule following machines"HN comment

  • "the slot machine can drop any hard requirement"HN comment

  • "I have production systems running that passed rigorous reviews"HN comment

  • "20x speedup in ability to ship"HN comment

  • "Proposal: Selectively Preserve Tiny, Scoped Repository-Memory Capsules"GitHub issue

  • "Drift data for 3 of these skills across the sonnet-4-6 to sonnet-5 release"GitHub issue

  • "npx packs only skills/"GitHub issue

GitHub45.0K

alibaba/open-code-review

Secure, fast, efficient, battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in multi-language ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible.

youtube
Open Code Review Tutorial: Alibaba Open-Sourced Their Internal Code Reviewer

Market Signal

Why It Has Market Pull

A real, actively maintained AI code-review tool open-sourced by Alibaba, backed by a dedicated site and a published benchmark; this is corporate-grade tooling with genuine builder and press momentum rather than a side project.

  • 45,113 GitHub stars and 3,259 forks, up from roughly 22,000 stars reported in press coverage about four months earlier
  • Reached #1 on GitHub Trending (around July 23, 2026) and drew 284 points / 73 comments on its Hacker News launch thread
  • Maintainers reported paired performance against Claude-4.6-Opus: 33.90% precision on their benchmark vs 7.23% for a generic coding agent on the same model, using 385K tokens per review vs 5,664K
  • 284 open issues currently tracked, including active community requests showing ongoing engagement
  • Covered independently by InfoQ and multiple technical blogs within months of its May 2026 release

feedbacks

What People Are Saying

  • "very good recall (~74%), not so good precision (~12%), F1 tanked (~20%)"HN comment

  • "Not working with gpt5.x models...max_tokens hardcoded"HN comment

  • "If 7/8 flagged items are fine, people ignore warnings"HN comment

  • "feat(allowlist): expand supported file types — language experts wanted"GitHub issue

  • "Wanted: Who is using Open Code Review? Please leave a comment!"GitHub issue

  • "divide and conquer strategy appreciated for combining deterministic checks with LLM flexibility"HN comment

GitHub28.1K

anthropics/knowledge-work-plugins

Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork

youtube
Claude Cowork Plugins Explained (& how to build AI employees)

Market Signal

Why It Has Market Pull

An official Anthropic repository extending Claude Cowork with role-based plugins for knowledge workers — this carries strong built-in credibility and real adoption signal purely by virtue of being first-party, and it has already accumulated substantial independent explainer coverage.

  • 28,180 GitHub stars and 3,241 forks, for a repo created in January 2026 — roughly 9 months old
  • Covers 15+ role-based plugin directions (engineering, sales, legal, finance, HR, operations, data analysis, biological research, etc.)
  • Actively maintained: last push on October 9, 2026, with 132 open issues currently tracked
  • Picked up by at least 6 independent explainer sites/newsletters within months of release
  • Apache-2.0 licensed, loadable in both Claude Cowork and Claude Code

feedbacks

What People Are Saying

  • "commands.md vs skill.md -- community guidelines"GitHub issue

  • "google-calendar MCP server turned down — productivity plugin broken"GitHub issue

  • "Plugins keep randomly disappearing"GitHub issue

  • "Built-in create-shortcut skill references non-existent set_scheduled_task tool"GitHub issue

  • "feat(health): Implement comprehensive healthcare organization plugin with conductor methodology"GitHub issue

  • "Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork"Dev.to article

  • "Independent third-party coverage is still thin."evidence gap

Product Hunt361

OpenSEO

Without good data, your agent gives generic advice. We provide the data, tools, and integrations your AI agent needs for SEO via MCP. Regular SEO is the foundation of good GEO, but we just released a new set of AI Visibility features as well.

youtube
OpenSEO - Open Source Semrush Alternative You Can Self-Host

Market Signal

Why It Has Market Pull

OpenSEO is a real, actively used open-source SEO platform (self-hosted alternative to Semrush/Ahrefs) with an MCP server for agent access; it combines a fast-growing GitHub repository, a credible low-cost pricing model, and specific, detailed user praise across multiple surfaces, making it one of the stronger picks in this set.

  • 22.9k GitHub stars and 3.0k forks on every-app/open-seo, with 729 commits on main.
  • 364 Product Hunt upvotes, ranked #2 product of the day on October 8, 2026.
  • A related Hacker News discussion cited 22,000+ GitHub stars gained in under 7 months.
  • Hosted pricing is $10/month with free self-hosting; a typical 50-keyword research session costs $1-5 in DataForSEO credits versus $99+/month for Ahrefs.
  • Two detailed, named Product Hunt reviews (5.0 average) specifically cite Claude Code and MCP integration as the reason for adoption.

feedbacks

What People Are Saying

  • "useful SEO tools without the upsell clutter"Product Hunt review

  • "an essential part of my web design process"Product Hunt review

  • "openseo is a godsend"Product Hunt comment

  • "22,000+ stars in less than 7 months"HN post

  • "Stop Paying $300/Month for SEO Tools: OpenSEO Is the Self-Hosted Secret"dev blog article

  • "Open Source Alternative to Semrush, Ahrefs and Ubersuggest"product directory listing

Product Hunt332

Claude Haiku 5.5

Claude Haiku 5.5 is Anthropic's fastest and most capable small model, built for high-volume, cost-sensitive tasks like summarization, classification, coding subagents, customer support, and browser use.

youtube
Claude Haiku 5.5 Is a GAME CHANGER! 90% CHEAPER & INSANE Performance! (Fully Tested)

Market Signal

Why It Has Market Pull

Claude Haiku 5.5 is a genuine flagship small-model release from an already-dominant AI lab rather than an unproven startup pitch, and it landed with real independent attention: a front-page Hacker News debate, a Product Hunt Launch-of-the-Day placement, and third-party benchmark coverage.

  • Hacker News discussion thread passed 300 points the day after the October 7, 2026 announcement.
  • Product Hunt named it Launch of the Day, ranking #3 across all launches on October 8, 2026, with 332 upvotes.
  • Artificial Analysis's Intelligence Index scored it 43 at max reasoning effort, ahead of GPT-6 Luna's 38.
  • List pricing is $0.10 / $0.50 per million input/output tokens under a 100k-token prompt, roughly 90% cheaper than Haiku 4.5 at that size.
  • Context window grows from 200k to 1M tokens and max output from 64k to 128k versus the prior Haiku release.

feedbacks

What People Are Saying

  • "overall very good at data analysis but falls short on deeper statistical questions"HN comment

  • "Qwen3.6 35B-A3B is both much faster and much more capable than Haiku"HN comment

  • "I'm more excited by the Haiku 5.5 announcement buried in this post"HN comment

  • "Haiku 5.5 is 40x cheaper than Opus 5.5, yet most people never change what their subagents run on"X reply

  • "benchmarks, real costs, and the 100k catch"tech blog review

  • "the compact and fast Claude Haiku 5.5 finally replaces 4.5"tech news article

Sources

GitHub

mattpocock/skills

282.4K

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

BerriAI/litellm

60.6K

The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]

cathrynlavery/diagram-design

47.6K

Editorial diagram design for Claude Code, Codex, GitHub Copilot, Factory Droid, and Pi. 42 diagram types. Self-contained HTML + SVG. No shadows. No Mermaid slop.

morluto/rea

40.8K

Reverse engineer anything with agents, from app behavior down to native binaries.

Robbyant/lingbot-map

17.6K

[ECCV 2026 Best Paper Award Candidate] LingBot-Map: Geometric Context Transformer for Streaming 3D Reconstruction

storytold/artcraft

10.7K

ArtCraft is an intentional crafting engine for artists, designers, and filmmakers

Product Hunt

OpenSwarm

380

Stop bolting AI onto your computer. OpenSwarm is an AI-first operating system where agents run the whole machine: apps create themselves, browsers control themselves, and swarms of agents work together. Your OS just became an infinite canvas, and your work just transformed from you doing one task at a time, to orchestrating dozens in parallel. Tell it what you want and a swarm of agents does the work with you at the helm.

Cekura Bench

169

Cekura Bench publishes voice AI benchmarks you can verify. Our new speech-to-speech benchmark tests 9 realtime voice models, including GPT Realtime 2.1, Gemini Live, Grok and Phonic, as complete phone agents on live calls: 82 scenarios, three runs each. Models are ranked on reliability, data accuracy, stalled calls, response time and cost, and every call transcript is public. Cekura Bench also covers voice agent benchmarks and STT benchmarks, with TTS benchmarks coming soon.

KloudMate 2.0

131

3 years after launching as an AWS serverless monitoring tool, KloudMate returns to PH as a full-stack, AI-powered Observability & Agentic SRE-Ops platform. It unifies metrics, logs, traces, and more, while KloudMate AI helps teams investigate incidents, find root causes, and resolve tickets. From dashboards to instant answers, KloudMate is built for modern, distributed systems. KloudMate AI Modules - Assistant (Answers), Builder (Dashboards, Alarms), Investigator (RCA), Docs (Documentation).

ChickyTutor

126

Hi Product Hunt! I’m Phanos, founder of ChickyTutor. We’re returning with an interactive whiteboard that brings visual teaching into spoken language practice. Ask about a language pattern, see examples, work through an exercise and discuss it with your tutor all during the conversation. We’ve also partnered with L’école de français to combine teacher-led French classes with daily AI practice between lessons at chickytutor.com/hybrid/french, and released ChickyTutor on Android.

Claude for Google Workspace

122

Claude now works directly inside Google Docs, Sheets, and Slides. Ask it to draft and edit documents, analyze spreadsheets, write formulas, build charts, or create presentation slides — all without leaving your file. You can also create and edit Google Workspace files directly from Claude using the new connectors. No more copying and pasting between apps. Available in beta for all paid Claude plans.

Polylane for Vercel

118

Keep shipping while Polylane watches your Vercel apps. Connect through the Vercel Marketplace to bring projects, deployments, domains and logs into one live graph. Agents catch regressions, investigate the root cause with evidence, and open a pull request with the permanent fix for your review.

YC Launch

Hacker News

Agent.reviews – Where AI agents read and write reviews on tools

71 pts

Hi HN! I’m Louis, Co-Founder of Armature (YC P26), where we help teams make their product discoverable and usable by coding agents. We already measured 50k+ agent sessions and realized that over and over agents would encounter the exact same limitations on different tasks using the same tool. So we wondered why these weren’t fixed. And the answer is simple: the feedback loop just doesn’t exist between agents and software vendors but also between different agents. Humans can share their experienc... (71 points, 49 comments).

Durable Actors – OSS Durable Objects with configurable compute

51 pts

Hi HN, we're Thomas and Olivier from Terse ( https://www.useterse.ai/ ) We've built Durable Actors, an open-source alternative to Cloudflare's Durable Objects. A Durable Object/Actor is a tiny server that handles one request at a time and has its own SQLite database. There's exactly one of each in the world and it is addressed by name. This is the perfect primitive for deploying multiplayer agents. Each agent can have its own Durable Actor, and each user can connect to that Actor via websocket.... (51 points, 25 comments).

Breadcrumb, record everything on your mac + context manager for AI

49 pts

Hi HN, I'm Justin. Breadcrumb records everything you do on your Mac (screen + meetings + AI transcripts + what you and your AI decided) and turns it into memory your AI can search. It's local and encrypted. You can also teach it rules by talking to it and it makes sure the right rules turn up in the right context. Works with Claude Code / Codex / Cursor / opencode. All of this is exposed to your AI as 30+ MCP tools (here's the definitions): https://innerloop.works/breadcrumb/mcp I started it in... (49 points, 9 comments).

HF Spaces

Qwen-Image-2.1

402 likes

Interactive demo for Qwen-Image-2.1 — unified text-to-image generation and image editing with native RGBA transparency support. 📑 Blog 🤗 Model Weights 💻 GitHub Qwen-Image-2.1 is a Hugging Face Space tagged with gradio, region:us. It has 402 likes on Hugging Face.

Viggle Turbo for Qwen-Image-2.1

233 likes

6-step Qwen-Image-2.1, T2I + editing, vs-base comparison Viggle Turbo v0.3 — 6-step Qwen-Image-2.1 A distilled Qwen-Image-2.1 that generates and edits images in 6 steps with no classifier-free guidance, about 5× faster than the 40-step base model. On most prompts it is hard to tell apart from the base model; small, dense text and complicated edits (multi-reference composition, face swaps, identity-preserving edits) can still fall short of it. v0.3 (2026-09-29): at 6 steps, less grain than v0.2.1 and a little softer on fine texture. We think 6 steps is close to its capacity: every further gain we found cost something elsewhere. The new 9-step setting runs 7 turbo steps and lets the base model...

MiniMax H3 Turbo LoRA

1.3K 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...

Jev Decision Index

550 likes

Benchmarks and news on various repros of TypeSafe's Jev Who is rebuilding TypeSafe's Jev (System One / RLCD) in the open? This static Space opens on the Decision Index leaderboard; the News tab tracks the artifacts in one combined grid, color-coded by kind: Decoding: parallel constrained decoding on stock models (inference technique, no new weights) Diffusion: text diffusion models run in a "Jev mode" Trained: Jev-like scoring heads and fine-tunes, weights often on the Hub, promised models listed last Prior art: "this already exists" claims Explainers: architecture speculation, explainers, benchmarks and roundups Cards sort by a trending score: ♥ likes on X + 5 × GitHub stars + 8 × Hub likes...

QIE-2511 Rapid-AIO LoRAs Fast (Experimental)

417 likes

Demo of the Collection of Qwen Image Edit LoRAs QIE-2511 Rapid-AIO LoRAs Fast (Experimental) is a Hugging Face Space tagged with gradio, mcp-server, region:us. It has 417 likes on Hugging Face.

Wan2.2 14B Preview

400 likes

generate a video from an image with a text prompt Wan2.2 14B Preview is a Hugging Face Space tagged with gradio, mcp-server, region:us. It has 400 likes on Hugging Face.