Skip to main content
Back

GitHub Copilot adds Kimi K2 , Claude auto-applies for jobs , pxpipe

Dev hacks Claude Code costs 70%, Mistral solves 587 Putnam problems, Sakana hits 93% on Sudoku, OpenMed ships on-device PII redaction at 755 tok/s

July 6, 2026 6 min read

Hey,

Kimi is everywhere today. GitHub Copilot just broke its closed-model streak by adding Kimi K2.7 as its first open-weight option, and Superpowers plugin is already wrapping it into structured agentic workflows.

The pattern is clear: open-weight models are landing inside the most mainstream dev tools on the planet. A legitimacy signal the closed-model camp should take seriously.

Meanwhile, Sakana's multi-agent paper quietly shows the real unlock: many small agents beat one giant brain. The stack is changing fast.
In Partnership with · partner_image
Author photo
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
Summary

Read time: 6 min 43 sec

Top Repo

▸ Dev turns a reMarkable tablet into a Harry Potter-style AI diary

ASUS

▸ ASUS AI POD: Built for Trillion-Parameter AI

Top Repo

▸ Open-source tool uses Claude to auto-apply for jobs, hits 3.5k GitHub stars

Slack

▸ 30 Ways to Automate Work in Slack

Top Repo

▸ Hack cuts Claude Code costs 70% by feeding context as images to OCR

Signals

1. OpenMed ships on-device clinical PII redaction model hitting 755 tokens per second on Mac
2. Presented by WebBrain: WebBrain: open-source AI browser agent. BYO model, runs locally/offline, MIT-licensed
3. New open-source format packages reusable skills directly into AI agents
4. Sakana AI's multi-agent framework hits 93% on Sudoku where baselines score 11%
5. DeepSeek releases R1, an open-source reasoning model on Hugging Face
6. Mistral releases math reasoning model that solves 587/672 Putnam problems at 10x lower cost
Top Repo
Dev turns a reMarkable tablet into a Harry Potter-style AI diary
15,314 Likes
Dev turns a reMarkable tablet into a Harry Potter-

Riddle turns a reMarkable Paper Pro tablet into Tom Riddle's diary from Harry Potter. You write with your pen, your words fade into the page, and an AI writes back in flowing handwriting, stroke by stroke. No screen, no keyboard, no chat box. Just ink on paper.

Here is what makes it tick:

  • After a 2.8-second pause, your handwriting gets sent as an image to a vision AI model (one that can read images), which reads it and replies
  • Works with any OpenAI-compatible API: OpenAI, Groq, OpenRouter, or even a local model
  • First reply appears in roughly 0.9 to 1.1 seconds, streaming sentence by sentence
  • Built in Rust, runs natively on the tablet

To try it, you need a reMarkable Paper Pro with SSH access enabled. Grab the prebuilt zip from the GitHub releases, copy it over with scp, drop in your API key, and tap launch. Still early, but fully open source under MIT.

Read more
forward →
Presented by ASUS
ASUS AI POD: Built for Trillion-Parameter AI

The ASUS AI POD with NVIDIA Vera Rubin NVL72 is a rack-scale AI system engineered for the next generation of AI reasoning.

The NVIDIA Vera Rubin Superchip delivers 10X NVFP4 inference performance per megawatt, 10X lower inference token cost, and 2.8X HBM4 memory bandwidth versus previous generations.

ASUS AIDC automates software cluster setup, enabling modern AI factories to be online within 30 minutes.

Trusted AI. Total Flexibility: ASUS AI POD with NVIDIA Vera Rubin NVL72

Get the full specs
partner with us →
Top Repo
Open-source tool uses Claude to auto-apply for jobs, hits 3.5k GitHub stars
7,000 Likes
Open-source tool uses Claude to auto-apply for job

Someone built a Claude agent to apply for jobs. And then open-sourced it. Meet ai-job-search, built by Mads Lorentzen, a PhD geophysicist who automated his own job hunt.

The idea is simple: you fill in your background once, point it at a job posting, and it handles everything. Here is what it does for you:

  • Reads the posting and scores how well you fit the role
  • Drafts a tailored CV, picking only your relevant experience
  • Writes a cover letter framed around what you bring to that specific role
  • A second AI agent reviews the first one's work, flags weaknesses, and the first agent revises
  • Compiles both into clean PDFs ready to send

To get started: fork the repo, run gh repo fork MadsLorentzen/ai-job-search --clone, add your profile, and fire away. You need an Anthropic API key or Claude Pro subscription plus Python 3.10+.

It hit 4.4k stars fast. MIT license. Every step is plain text you can read and change.

Read more
forward →
Presented by Slack
30 Ways to Automate Work in Slack

Your team is spending time on work that shouldn't require human attention - status updates, approvals, follow-ups, and handoffs that repeat day after day. 

Slack automation changes that. From auto-routing requests to triggering cross-tool workflows, this guide covers 30 practical ways teams are cutting manual work and reclaiming focus.

Get the guide
partner with us →
Top Repo
Hack cuts Claude Code costs 70% by feeding context as images to OCR
4,844 Likes
Hack cuts Claude Code costs 70% by feeding context

pxpipe is a local proxy that cuts your Claude Code API bill by converting bulky text context into images before sending it to the model. Here is the core trick: images are priced by pixel size, not by how much text is packed inside. Dense code and JSON fits roughly 3x more content per token as an image than as plain text.

The result? A real 59-70% cost reduction end-to-end on production traffic. Try it in 30 seconds:

  • Run npx pxpipe-proxy to start the proxy locally
  • Point Claude Code at it: ANTHROPIC_BASE_URL=http://127.0.0.1:47821 claude
  • A dashboard shows live token savings and every text-to-image conversion

But here is the catch you need to know: it is lossy. Exact strings like IDs, hashes, and secrets can be silently misread. Fable 5 reads hex strings correctly 13 out of 15 times. Opus gets 0 out of 15. Keep byte-sensitive values as text, which pxpipe does automatically for recent turns.

Read more
forward →
Signals
1
OpenMed ships on-device clinical PII redaction model hitting 755 tokens per second on Mac
1,494 Likes
2
WebBrain: open-source AI browser agent. BYO model, runs locally/offline, MIT-licensed
Presented by WebBrain
3
New open-source format packages reusable skills directly into AI agents
22,405 Stars
4
Sakana AI's multi-agent framework hits 93% on Sudoku where baselines score 11%
701 Likes
5
DeepSeek releases R1, an open-source reasoning model on Hugging Face
8,080,000 Downloads
6
Mistral releases math reasoning model that solves 587/672 Putnam problems at 10x lower cost
1,423 Likes
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 200,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.

Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
Looking to promote your company, product, service, or event to 250,000+ AI developers?
Work with us