What we call "age verification" is actually mass surveillance

https://i0.wp.com/craphound.com/images/23Jun2026.jpg?w=840&ssl=1
The user discusses online surveillance, age verification, and its implications on kids and privacy, arguing that it's a surveillance industry dream. They also mention various books and events, including their own works, and a call to action to prioritize privacy over age verification.

F3

https://opengraph.githubassets.com/4ba42fe33148f1fcb1b4462a53a442bba5eb3919387bc9c211c2a1ee8d03606c/future-file-format/F3
F3 is a next-generation data file format designed for efficiency, interoperability, and extensibility. It embeds Wasm decoders to ensure compatibility on any platform.

Unlimited OCR: One-shot long-horizon parsing

https://opengraph.githubassets.com/73a749dfc147a1186cf79faa0752f07e22596d3af6394fd880eea4f2729dba11/baidu/Unlimited-OCR
The model Unlimited-OCR is available on ModelScope for inference using Huggingface transformers on NVIDIA GPUs. It supports single image and multi-page parsing with various configurations.

Mistral OCR 4

https://mistral.ai/images/heros/hero-ocr4-image.jpg
Mistral OCR 4 is a document-understanding model that extracts text, bounding boxes, block types, and confidence scores from 170 languages. It's available through a single API endpoint and can be used for enterprise search, RAG, and agentic workflows.

Israel targeted Gaza children resulting in genocide, UN inquiry says

Please enable JS and disable any ad blocker

Crypto in 2026: Oh, This Is the Bad Place

https://www.stephendiehl.com/images/sd-social-preview.png
The crypto industry has created a high-throughput onboarding ramp for retail gambling, preying on young people's economic anxiety and financial nihilism. The industry's products, including prediction markets and stablecoins, are essentially self-referential games that lack any real-world value or information.

VibeThinker: 3B param model that beats Opus 4.5 on reasoning with novel SFT+GRPO

https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png
VibeThinker-3B achieves frontier-level performance on verifiable tasks with 94.3% on AIME26 and 96.1% on LeetCode contests. It matches larger models like DeepSeek V3.2 and GLM-5 in performance, suggesting compact models can be a path to frontier-level performance.

Jerry's Map

https://images.squarespace-cdn.com/content/v1/551565dce4b057e153ef5fd7/1449864504674-YK78OW5V8Z0MRJT3WSC6/image-asset.jpeg
Jerry created a 4000-panel map project using a custom deck of cards with instructions for each panel. The project is a two-dimensional "virtual world" art project with a set of rules and randomly generated instructions.

Show HN: TikZ Editor – WYSIWYG editor for figures in LaTeX

https://tikz.dev/editor/assets/app-screenshot-2400-CXE6z4gU.png
The app allows editing TikZ figures with drag-and-drop functionality and instant code updates. It supports common TikZ features, including \foreach loops, and has a free, open-source license.

The Coming Loop

https://lucumr.pocoo.org/social/2026-06-23-the-coming-loop-social.png
The user is uneasy about the future of software development where loops and AI assistants dominate, potentially removing human responsibility and understanding. They question how to retain good engineering rules and human supervision in a future where loops and machines increasingly drive code development.

Fired by Google for creating the Google workspace CLI

https://pbs.twimg.com/media/HLhHJvWaoAEBgxd.jpg
I was fired by Google two months ago for creating a viral Google Workspace CLI that disrupted their plans. Google announced an official Workspace CLI two days before my termination, revealing their fear of disruption.

FUTO Swipe – A new swipe typing model

https://swipe.futo.tech/hero-2.webp
The Encoder model makes swipe typing predictions, but lacks accuracy, while the ContextLM model improves predictions by eliminating nonsensical words. Combining all 3 models with a beam width of 300 achieves a top-4 fail rate of ~4% and an error rate below 1% on the test set.

Will It Mythos?

https://swelljoe.com/img/full-bench.png
The author created a benchmark suite to test various AI models' ability to find security bugs, specifically those found by Mythos, and found that some models, like Gemma 4 and Qwen 3.6, performed well.

Madison Square Garden compiled a list of activists against facial recognition

https://storage.ghost.io/c/0f/76/0f76b548-bc58-4f25-abc3-3f5ebca07da4/content/images/2023/08/404-joseph-01-1.jpg
Madison Square Garden compiled a list of activists who criticized its facial recognition technology, including Evan Greer, Adam Schwartz, and Albert Fox Cahn. The document, found in a data breach, included background information, contact details, and quotes from the activists.

AI's Affordability Crisis

https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgPaajVCJYuSt8Rkj3fR8GSZQ_pAbHJldOsS17qRHqBrH8Ps-tDNb1l1sLmgZWOC6s1-fs5EpZhIJzc5lwWLSJz2tNfu9gvvG9WSU_2uVwaxIBG39ZiRj1Lthyphenhyphenv_y8qBVFlYIdDQvoNpm6Nj8KODEmw0lrbKuHcTs7zVDWIyNOo3cu88-dWjXWS_T87fbvp/w200-h111/Subsidy.png
AI platforms are subsidizing their products to generate massive demand, but this strategy is unsustainable as costs are rising and users are being forced to switch to token-based pricing, which is eye-wateringly expensive. The AI industry is facing a debt crisis, with estimated debt of $3T, and needs to replace human labor on a colossal scale to service the debt, which is a daunting task.

In praise of memcached

The user is discussing the misuse of Redis as a database due to its ease of use and features, leading to maintenance and monitoring issues. They prefer Memcached for caching due to its simplicity and stateless architecture, making it easier to handle downtime and clustering.

California AB 2047 makes 3D printers off-limits to students, educators, business

https://www.the3dprintingnerd.com/ab2047-letter.png
Send a single email or make a direct call to urge a NO vote on AB 2047, citing its rushed passage and contradictions, as well as its unworkable technology requirements and unfair penalties. The bill also unfairly targets small businesses and makers, while exempting undefined groups such as entertainment studios.

1,700 free online courses from top universities

https://www.openculture.com/wp-content/themes/openculture_v4c/images/open_culture_white.jpg
If you take the course for a certificate/credential, you’ll be charged. we will receive com­mis­sion from our affil­i­ate partners–coursera, future­learn and edx. the u.s. gov­ern­ment re­quires that all students have id and passwords to access the courses. you must have access to the internet to

Claude Tag

https://www.anthropic.com/_next/image?url=https%3A%2F%2Fcdn.sanity.io%2Fimages%2F4zrzovbb%2Fwebsite%2F1cbbe8ed6d2b55d5f27deaf44216f7e2b8ed29b0-1920x1080.png&w=3840&q=75
Claude Tag is a new feature that allows teams to tag @Claude in Slack for tasks, and it builds context by remembering relevant information. It's available in beta for Claude Enterprise and Team customers, with plans to expand availability.

The worthlessness of Vitamin D is mildly exaggerated

https://dynomight.net/img/vitamin-d/ap74.37.jpg
The user discusses the benefits and limitations of vitamin D supplements, citing various studies and trials that show mixed results, but concludes that taking vitamin D may be beneficial for people with low levels, particularly those with severe deficiency. The user also suggests that daily doses of vitamin D may be safer and more effective than large, intermittent doses, and that raising ...

The new HTTP QUERY method explained

https://kreya.app/blogposts/new-http-query-method/query-example.png
A new HTTP method, QUERY, has been introduced to handle complex search queries with a request body, replacing the need for POST in read-only requests. QUERY is similar to GET but with a body, intended to be safe and idempotent, allowing for caching and retrying failed requests.

OpenAI DayBreak – GPT-5.5-Cyber

https://images.ctfassets.net/kftzwdyauwt9/7x54Eb1FniAlhNjopx6bKr/251c02c82c39cb8f82ca2f8401f65eac/Inline_Stats__1_.png?w=3840&q=90&fm=webp
OpenAI is expanding Daybreak to help organizations patch vulnerable software at machine speed, using models like GPT-5.5-Cyber and Codex Security.

Elevated error rate across multiple models

Claude's Status Page - Elevated error rate across multiple models.

In memory of the man who put red and green squiggles under words

https://devblogs.microsoft.com/oldnewthing/wp-content/uploads/sites/38/2019/02/ShowCover.jpg
Tony Krueger worked on early versions of Microsoft Word and introduced the unobtrusive spell checker with red squiggles under misspelled words. His feature became a standard in word processors and was even referenced in a "Weird Al" Yankovic parody video.

Extreme Heat conference cancelled due to extreme heat warning

https://www.lse.ac.uk/granthaminstitute/wp-content/uploads/2025/07/London_heat.jpg
The London Climate Action Week event has been cancelled due to extreme heat warning. It was to announce the Adeline Stuart-Watt Award winner and discuss extreme heat governance globally.

75% More Pedestrians Have Been Killed Since 2009. Giant Trucks and SUVs Are Why

https://www.thedrive.com/wp-content/uploads/2026/06/SUPERDUTY.jpeg?quality=85&w=1920
Pedestrian fatalities in the US have increased by 75% since 2009 due to larger vehicles. Cars growing in size and weight have led to thousands of preventable deaths over the past 16 years.

Anthropic updates their terms to verify age or identity

https://cdn.sanity.io/images/4zrzovbb/website/6d4a0d28992ade92d6fa63646fd9c9d318245c6c-2400x1260.jpg
Anthropic collects personal data from users of its website and services, including names, email addresses, and payment information, to provide and improve its products and services.

Swift Package Index joins Apple

https://swiftpackageindex.com/images/logo.png
Swift Package Index has joined Apple, preserving its vision and expertise to build a comprehensive package registry for the Swift community. The registry will continue to support package discovery, security, and reliability, with new features and capabilities to be introduced over time.

Digital euro clears key hurdle as EU seeks to break free from U.S. credit cards

https://s.yimg.com/ny/api/res/1.2/sSFgDee.bzvjHCfkl4ZWcA--/YXBwaWQ9aGlnaGxhbmRlcjt3PTk2MDtoPTY0MA--/https://media.zenfs.com/en/reuters-finance.com/433caafdaded4d5d25db94317076a334
The European Central Bank secured key backing for a digital euro, an electronic means of payment to reduce reliance on US credit cards. The digital euro will allow euro zone residents to make online and in-person payments, with a planned launch in 2029.

Polymarket's viral videos showed people winning big, but the bets were fake

https://cdn.arstechnica.net/wp-content/uploads/2026/06/polymarket-640x427.jpg
Polymarket paid social media users to create fake bet videos to promote its unregulated platform. The company's campaign racked up over 140 million views on TikTok, YouTube, and Instagram.

Don't verify email addresses by sending spam to them

People tried to validate email addresses by sending spam, retrying from different servers, but this approach is flawed and can be easily blocked by content filtering. This method is likely facilitated by a SaaS or automated tool, possibly created by a misconfigured or malfunctioning AI.

Trains halted across Germany because of communication system problem

https://dims.apnews.com/dims4/default/94c503b/2147483647/strip/true/crop/640x236+0+0/resize/320x118!/quality/90/?url=https%3A%2F%2Fassets.apnews.com%2Fc3%2F4c%2F65482a7b452db66043542c093eaf%2Fpromo-2x.png
A technical issue with Germany's GSM-R communication system halted all trains nationwide, leaving passengers stranded. Deutsche Bahn is working to resolve the issue and provide assistance to affected passengers.

San Diego photologs from the 1970s

https://storage.ghost.io/c/7e/ed/7eed3692-c4ce-47fb-af77-6293d8e19002/content/images/2026/06/san-diego-header-1.png
A collection of high-resolution photolog scans from San Diego in the 1970s reveals a colorful world of pastel-colored cars and whimsical signage. The footage has a distinctly Wes Anderson vibe, with a warm, sun-washed palette and creative typography.

Wikipedia cofounder Larry Sanger blocked from editing Wikipedia

https://upload.wikimedia.org/wikipedia/commons/thumb/6/6f/Larry-Sanger-headshot-768x1024.jpg/120px-Larry-Sanger-headshot-768x1024.jpg
A co-founder of Wikipedia has been banned and is requesting review of the banning policy. They aim to reform Wikipedia by promoting intellectual diversity and recognizing quality contributions.

Algorithmic Monocultures in Hiring

https://hai.stanford.edu/_next/image?url=https%3A%2F%2Fhai.stanford.edu%2Fassets%2Fimages%2Fhiring-rejection-rates.png&w=3840&q=100
AI hiring tools increase racial bias and shut out certain groups from jobs, with 26% of Black applicants and 15% of Asian applicants facing discrimination. Market concentration and shared dependence on a single vendor exacerbate these issues, highlighting the need for independent research into algorithmic hiring.

The Low-Tech AI of Elden Ring

FROMSOFT uses a low-tech AI system in their Soulsborne games, based on a stack of states called Goals, which can be updated and modified to create complex behavior. The system uses a mix of Lua scripting and C++ code, with a focus on flexibility and ease of use for designers and animators, making it well-suited for creating diverse and punishing NPC encounters.

Vulnerability reports are not special anymore

https://assets.buttondown.email/images/1ec4ddf5-598f-4d72-8d55-10dc1328aa76.jpeg?w=960&fit=max
As an open source maintainer, it's essential to view every issue, PR, and feedback as a present, not an obligation, except for vulnerability reports which require responsiveness and attribution due to the service provided by security researchers. With the rise of LLMs, the insight and confidentiality provided by security researchers are no longer scarce, and the focus should shift to triage, ...

United Wizards of the Coast recognized by NLRB

Workers at Arena's UWOTC-CWA union have been elected after months of planning and organizing. The union thanks supporters for their help and hopes their victory inspires others to unionize.

Lift4D: Harmonizing Single-View 3D Estimation for 4D Reconstruction In-the-Wild

https://lift4d.github.io/static/methodology/occ_mask_0.png
Lift4D is a test-time optimization framework for 3D reconstruction from monocular video, combining visual cues with data-driven priors. It improves over prior methods by providing coherent initialization and occlusion-aware optimization for challenging in-the-wild sequences.

Oracle shed about 20k roles globally in the last year

https://ichef.bbci.co.uk/news/480/cpsprodpb/011d/live/21b2c920-6e9a-11f1-ae80-bf13157382fa.jpg.webp
Oracle cut 21,000 roles globally in the last year due to AI reshaping its business. The company had 141,000 employees as of May 2026, down 13% from the previous year.

Apple is going to raise device prices, but when?

Mark Gurman thinks Apple will raise prices in the fall, starting with iPhones 18 Pro and Ultra in September, due to the global RAM shortage. He believes Apple may wait for new hardware to roll out price increases, rather than mid-cycle.

The truth about being a manager

https://sofiakodar.github.io/posts/becomingmanager/becoming_manager.png
Being an engineering manager is a challenging and lonely journey where you'll need to manage stress, learn to be quieter, and handle difficult conversations. You'll need to be professional, diplomatic, and a calm leader who drives progress and earns respect.

An Introduction to YOLO26

https://storage.ghost.io/c/2c/8d/2c8d8c0d-1c15-4b6d-825e-02b78d61d40a/content/images/size/w1200/2026/01/rfdetr-seg--1-.png
YOLO26 is a multi-task model family for object detection, segmentation, and classification, optimized for edge deployment with faster CPU inference and lower latency. It features five size variants and improved training with the MuSGD optimizer, outperforming previous YOLO generations in edge deployment.

AI Built a Nuke and Still Lost

https://www.lwilko.com/_next/image?url=%2Fblog%2Fcivbench%2Fthis_is_fine.jpg&w=3840&q=75
The author built a benchmark called CivBench to measure the strategic competence of AI models in a game of Civilization VI, and found that they consistently failed to notice important information and execute their plans due to the "sensorium effect" and the "knowing-doing gap". The benchmark revealed that AI models are prone to opportunism and pragmatic decision-making, but also showed a lack ...

Mark Zuckerberg directed Meta to create a prediction markets app

Please enable JS and disable any ad blocker

Show HN: Neural Particle Automata

https://selforg-npa.github.io/data/images/teaser.jpg
SPH perception aggregates nearby particles using smooth kernels to estimate quantities like density and state gradients. It preserves locality in Neural Cellular Automata while allowing particles to live on irregular configurations.

SpaceX sheds $400B in market value as debut rally hits reverse

The following information can help our support team to resolve this issue.

All train services in Germany halted after train radio communications disruption

Flash: All train services in Germany halted after train radio communications disruption: media-

Ask HN: Anthropic banned me from using Claude Code and I don't know what to do

User was banned from Claude Code twice, first due to VPN usage and second due to credit card, despite contacting support. User seeks visibility on their case through HN to resolve the issue.
The user discusses the importance of GIF fallbacks for old browsers and explores image compression techniques, including ZopfliPNG and flexiGIF, but finds that flexiGIF can give worse results due to its greedy approach. The user then implements their own GIF compression tool, ZGIF, which uses a hybrid search with pruning to achieve faster results while still beating the current state-of-the-art.

Show HN: Shumai – open-source Frame.io alternative for creative work

https://raw.githubusercontent.com/shumaiOne/shumai/main/docs/screenshot.webp
Shumai is a media management tool with features like secure storage, frame-by-frame annotations, and AI-powered metadata autofill. It can be run locally or remotely with Docker and Docker Compose.

Show HN: Bun-sqlgen – Type-safe raw SQL for Bun, no ORM

https://opengraph.githubassets.com/8af541211ab93a9ce4009a7bc64cd60873acb04b8b929aeaffcec21bc58bee8a/ilbertt/bun-sqlgen
The bun-sqlgen library generates typed SQL queries at compile-time, catching errors before runtime. It uses migrations as the source of truth for the schema and writes generated types to a file for static type checking.

The Traditional Vi (2007)

The vi editor was developed in 1976 by Bill Joy at UCB, but its source code was restricted until 2002 when Caldera removed usage restrictions. This port of vi is a free version with international character support and minor enhancements.

Ultralytics YOLO26: Unified Real-Time End-to-End Vision Models

https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png
Ultralytics YOLO26 is a unified real-time vision model family addressing limitations of YOLO detectors through coordinated architecture and training advances. It achieves 40.9-57.5 mAP on COCO at 1.7-11.8 ms latency, advancing the accuracy-latency Pareto front.

Pact: Anonymous Credentials for the Web

https://hacks.mozilla.org/wp-content/uploads/2026/06/pact-drawings-1.excalidraw1-scaled.png
The current anti-abuse systems on the web are failing to distinguish between humans and bots, leading to increased friction and compromised privacy for users. A proposed solution, PACT, aims to provide a privacy-preserving rate limiting system that uses scarcity signals from various sources, such as VPN providers, to enable websites to rate limit users without compromising their privacy.

I am a person who will look at the Steam Machine and cry

Valve's new Steam Machine is priced too high at $1400, making it unaffordable for many. The user will instead use their Steam Deck and a new TV to play low-power games.

Show HN: A pure ARM64 Assembly web server, now on Linux with CGI for no reason

https://raw.githubusercontent.com/imtomt/ymawky/linux/docs/ymawky.png
ymawky is a hand-written ARM64 assembly web server for Linux with syscall-only functionality and no libc. It supports static file serving, CGI scripts, and custom error pages.

80386 Early Start Memory Access

https://nand2mario.github.io/posts/2026/80386_early_start/timings.png
The author of the z386 FPGA core added the 80386's Early Start feature, which overlaps the memory access of the next instruction with the last cycle of the current instruction, resulting in a 39% increase in Doom performance and reaching ao486-class performance. The author also implemented other optimizations, including a store queue, cache geometry changes, and a wider and shallower ...

Prairieland Defendants Sentenced Today to Prison Terms Ranging from 30-100 Years

https://prairielanddefendants.com/wp-content/uploads/2026/01/532433083_17851813179523571_3910293510227401041_n.jpeg
Eight Prairieland defendants were sentenced to 30-100 years in prison for federal charges related to a July 2025 protest at an ICE detention center in Texas. Family and supporters vow to continue fighting for their freedom, calling the sentences cruel and disproportionate.

Show HN: Treedocs: Documentation that automatically checks for staleness

Treedocs is a tool that provides a visual overview of a repository, highlighting issues with red and green colors. It helps maintain documentation with automatic drift checks and version-controlled summaries in treedocs.yaml.

Who Does What? Team Topologies for the Agentic Platform

https://blog.owulveryck.info/assets/team-topologies-agentique/octo_team_topologies_responsabilites.en.svg
The agentic platform absorbs technical complexity, allowing business teams to focus on domain expertise and reducing cognitive load. Team Topologies is adapted to define four team types and three interaction modes to distribute and regulate cognitive load in the agentic production process.

You already have a Git server

https://gopeek-lovat.vercel.app/GoPeek_Logo.png
You can use SSH and git to work on server-side files locally, then push changes without SSH lag or errors. Git can also run shell scripts on new pushes, sending output to your terminal for immediate feedback.

ATProto Permissioned Data Proposal Draft

https://opengraph.githubassets.com/442cf9efee61fd336e832c5a8a6da78eba879d6717d25bf248c1570c2bac5ede/bluesky-social/proposals/pull/94
The batch cannot be applied due to multiple invalid suggestions. Suggestions cannot be applied in certain situations such as deleted lines, pending reviews, and queued pull requests.

Matrix and Quaternion FAQ

Matrix and quaternions faq - how do i calculate the inverse of a 3x3 matrix? -1 m is 0 m generates the identity matrix 1 m leaves the matrix unchanged. negative the rotation angle is equivalent to generating the transpose of matrix xyz = | || | yy = z o shears x by y | sij implements an axes shearing of i by

Let's say you're hungry for a sandwich

https://media.beige.party/site_uploads/files/000/000/003/@1x/724bb0d304e9c99c.png
A teacher explains a scenario where a deli owner forces customers to buy expensive sandwiches, then sells them at a lower price to employees, making the owner a trillionaire. This is compared to Elon Musk's SpaceX IPO, where the public is forced to buy stock at a high price, then employees sell at a lower price, making Elon a trillionaire through a similar pump and dump scheme.

Scars mark Britain's economy 10 years after Brexit vote

Please enable JS and disable any ad blocker

A Man Was Gifted His Dream Car by the Notorious Hacker He Put in Prison

https://www.thedrive.com/wp-content/uploads/2026/06/nun-hero.jpg?quality=85&w=1920
Kevin Mitnick, a hacker-turned-security consultant, was caught by Shawn Nunley, a Novell network administrator, after Nunley recorded a suspicious voicemail from Mitnick. Mitnick later apologized to Nunley and they became friends, with Mitnick leaving Nunley a gift to buy his dream car after his passing.

The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A" (2023)

https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png
Large language models fail to generalize from "A is B" to "B is A" due to the Reversal Curse. This limitation is observed across various models and sizes, including GPT-3 and ChatGPT.

Audit finds San Francisco tax official steered $10M contract to friend

https://assets.sfstandard.com/image/994911177489/image_9bd615fsm521f0hn57rofstb7h/-C2561x2561%2C0%2C428-S3840x3840-FPNG
San Francisco's chief assistant treasurer steered a $10 million contract to a tech company led by her friends, violating city ethics and procurement policies. An audit found the treasurer abused her position and created a "pay-to-play" business tax system, prompting the treasurer's office to reorganize its leadership.

WaPo Loves Data Centers More Than Disclosing Bezos's Financial Interest in Them

https://washingtonian.com/wp-content/uploads/2026/02/Screenshot-2026-02-11-at-33218-PM.png
The Washington Post's opinion section frequently promotes data centers without mentioning its billionaire owner Jeff Bezos' business interests in the topic. The paper's news side, however, has taken a more skeptical approach to data centers, documenting their downsides and Bezos' connection to the issue.

The End of Code Review: Coding Agents Supersede Human Inspection

https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png
Coding agents can serve all code review goals at lower cost and higher throughput, making human review unnecessary. Traditional human code review is a dead end when integrated with AI-assisted coding.
https://modal-cdn.com/blog/images/introducing-auto-endpoints-comparison-diagram.webp
Modal allows teams to own their inference without compromising on cost-performance or developer velocity. It provides a production-ready service with transparent code and metrics for debugging inference issues.

Show HN: Ideate a trading strategy with an Ex-Citadel Trader

https://sean-but-ai.vercel.app/me.jpg
The user worked at Citadel Securities for four years, making markets in single-stock options and managing risk. They now help traders build robust strategies by analyzing their ideas through a market-making desk perspective.

Backblaze Announces Five-Year Multi-Exabyte Agreement with CoreWeave

You don't have permission to access "http://www.businesswire.com/news/home/20260622788628/en/Backblaze-Announces-Five-Year-Multi-Exabyte-Data-Storage-Agreement-with-CoreWeave" on this server.

Show HN: The Cascade Graph – An interactive map of AI and energy constraints

https://atomprophet.io/images/logo.png
The Cascade Graph is a knowledge map of the physical economy, showing how stress flows through the real economy. It identifies drivers, chokepoints, and investable tickers to help find where pain cascades and what to own on both problem and solution sides.

I automated my job (and it made me a better leader)

https://github.blog/wp-content/uploads/2026/06/MSFTBuild_Blog_Header_01.jpg?resize=1600%2C850
The author, a senior director at GitHub, uses automations in the GitHub Copilot app to streamline her workflow and reduce stress. The automations connect her work across multiple platforms and tools, freeing up her time for more important tasks.

Show HN: RLM-based local debugger for AI agent traces

https://raw.githubusercontent.com/context-labs/HALO/main/assets/halo-icon.png
HALO is a methodology for building recursively self-improving agent harnesses using RLMs, which identifies issues in production agent deployments and suggests fixes. It uses a specialized form of RLM and can be installed via CLI or Python SDK, with optional telemetry and observability features.

Gemini models increasingly stucking in thinking loop

Sorry.