Tools

Author @0xSero recommends sitegeist (2026-06-16), calling it a cleaner alternative to parchi and noting they are actively porting features such as record and custom providers into their fork.

Origin, announced 2026-06-16 by Cursor/Graphite engineer Tomas Reimers (shared via @swyx), is presented as a Git alternative optimized for agent-scale workflows.

Cursor announced Origin, a code storage and git hosting product that "gives teams and agents a place to host, review, and collaborate on code." The June 16, 2026 post says Origin will be available this fall and directs interested users to join the waitlist at cursor.com/origin-waitlist.

Fable's Library of Babel implementation, completed posthumously by Opus 4.8, recreates Borges's concept and is described as quite accurate.

Ethan Mollick posted that he prompted Opus 4.6 (Claude Code) to generate Borges’s Library of Babel; within roughly an hour it produced a searchable web implementation that encodes books using a Feistel cipher so every possible book has a real location.

Coinbase introduced 'Coinbase Advisor,' an AI agent it says it registered with the SEC as an investment advisor on 2026-06-16.

text-to-lottie v1.0.0 is a stable release from @konstipaulus that creates production-ready Lottie animations by leveraging Claude Code and Codex.

OpenAI announced the OpenAI Developers plugin for Codex on 2026-06-15, a tool designed to speed development by guiding users through API key setup, surfacing the right OpenAI docs, and assisting with debugging steps.

The post by @jxnlco (published 2026-06-15) invites followers to guess what they are working on and links to a GitHub repository (github.com/jxnl/personal-mon…).

@viharkurama warns that handing real work to AI agents requires centralized knowledge: agents are limited by the docs they can reach.

Lead times for transformers are the topic: ZeroHedge asserts data-center developers face 2.5-year waits for power transformers and 3-year waits for step-up units (June 15, 2026).

The Bored Room portrays an eight-hour waiting game where stalled approvals, time-limited banking hours, and FX slippage caused a payment to fail.

Arc is Circle’s new platform introduced on 2026-06-15 that the company bills as an economic operating system: a compute environment to store value, move money, create corporate entities, and intermediate complex contracts.

PostHog announced Replay Vision on 2026-06-15 as a tool that automates reviewing Session Replay recordings.

hn_local_image, maintained by @ivanfioravanti, was updated on 2026-06-15 to include mlx-vlm and three new image models—Gemma 4 E4B, ernie-image-turbo, and ideogram-4-fp8.

hn_local_image v0.3.2 is a Local AI release offering quick CLI workflows (generate to eink/web and compare with style/watermark flags), a skill for agents, gemma-4-e4b for Hacker News analysis, five image models, and automatic publishing to @useTRMNL.

ZCode announced ZCode 3.0 running on GLM-5.2 for GLM Coding Plan users, emphasizing improved Agent task execution, long-context coding, and a new Goal feature to manage large development objectives.

A short exchange promoting a distributed-compute project for Apple Silicon: @ivanfioravanti says they love the idea and will try it tomorrow (post dated 2026-06-15).

User @usreneray tweeted on 2026-06-14 that they have once more become enamored with a nonliving object — a 1940s 'Saturn Ceiling Light' promoted by Spacesthetic (@interiorsuckerr).

The X post from @interiorsuckerr (June 15, 2026) is a short video showcasing a double-faced clock, described by the author as having "so much charm." The clip emphasizes the clock's dual-faced design and decorative appeal, targeting viewers interested in statement or vintage-style home decor.

Arthur Vallin's oak "Chubby" coffee table is being admired: on 2026-06-14 @ImJamesBaldwin reshared Spacesthetic (@interiorsuckerr)'s post with emphatic praise (emojis and "What a Beaut").

AlexBellamyLeto reposts Spacesthetic (@interiorsuckerr) on 2026-06-15, captioning an image of MOKY — a 2026 design by Davide Bozzo — with the emphatic line “Si es funcional, la necesito al 100%,” signaling strong preference for functional design and endorsement of Bozzo’s piece; the original is linked from Spacesthetic’s post.

Ian Miles frames Factory's launch as a practical, production-focused take on the Factorio concept: Factory 2.0 (announced 2026-06-15) shifts the product narrative 'from coding agents to software factories,' explicitly calling the approach a 'Software Factory' and linking to a short announcement video.

CBum upgraded his home gym with new gear showcased on 2026-06-15 by @homegymcoop, emphasizing novel 'barbell collars of the future' created through a Sorinex–Gungnir partnership.

NVIDIA's build.nvidia.com is presented as a free AI API platform (published 2026-06-15) hosting 100+ models across chat, code, vision, and reasoning, with no billing setup or credit-card requirement and non-expiring credits.

FactoryAI announced Factory 2.0, a platform it calls an autonomous "software factory" that automates the entire SDLC, adapts from external signals, and self-improves.

Hamptonism says they no longer write code, instead spending their time triaging, reviewing, and making sure work ships.

Patty describes a private Tokyo wellness rental priced at ¥54,000 (~$336) that gives sole access to an onsen, cold plunge, and sauna.

Author @godofprompt praised OpenRouter’s June 13, 2026 release, saying their work merely builds on OpenRouter and calling the launch “priceless.” OpenRouter unveiled the Fusion API, billed as the market’s smartest compound model that matches Fable-level intelligence while cutting cost by half, and teased a follow-up explanation of how Fusion works.

Author @godofprompt says an upcoming open-weight model — released this week under the MIT license — offers a 1M-token context and costs about one-tenth of Fable 5.

Gemini Omni generates clean 10-second videos from one reference image or other inputs (text, photo, audio).

EdrawMax is presented as an instant diagram tool that converts one prompt into a fully editable diagram in 4 seconds, supporting more than 210 diagram types.

EdrawMax, demonstrated by @ginacostag_ on 2026-06-15, can produce a full product-launch slide from one prompt—layout, sections and visuals generated instantly—and crucially lets users edit every single element afterward.

The post by @ginacostag_ (2026-06-15) demos an "AI Diagram Agent" that converts one text prompt into over 210 diagram types.

@ginacostag_ showcased a new 2D→3D floor‑plan tool on 2026-06-15, highlighting that after building a 2D plan you can press one button to produce an immediate 3D render.

A single tool, described by @ginacostag_ on 2026-06-15, generates images from simple descriptions (posters, org charts, engineering graphics) and offers integrated editing like background and object removal so no app-switching is required; the author also notes the tool handles video.

A June 15, 2026 post from @ginacostag_ points out a feature-rich diagramming toolkit—over 30,000 templates, drag-and-drop symbols, Visio import/export, and PDF/PNG/PPT output—that has driven quiet adoption among engineers, product managers, and educators who need rapid diagram creation and cross‑tool compatibility.

Gina Costag (@ginacostag_) argues on 2026-06-15 that AI can reclaim the time people waste building diagrams manually and promotes EdrawMax as a free-to-start solution (link: cutt.ly/vt2Gfr0I).

Cartesia announced a "full stack" voice AI milestone on 2026-06-15: Sonic-3.5 (streaming text-to-speech) and Ink-2 (streaming speech-to-text) are claimed as the #1 streaming models for speaking and listening, respectively.

User @ginacostag_ reports that Tripo converted a single uploaded photo into a complete 3D model in under a minute, producing clean topology that could be dropped straight into Blender.

Tripo's 1st Anniversary special is live (post dated 2026-06-15): new users get 500 bonus credits, Pro costs $9.9 for the first month, and promotions can deliver up to 50% off with double credits—discounts apply automatically without a promo code.

Gina Costag (@ginacostag_) touts a pipeline (post dated 2026-06-15) that starts by generating an image from text or a reference photo and produces a clean, 3D-ready asset.

Then Smart Mesh is a fast mesh generator that produces clean, low-poly base meshes in roughly five seconds.

The post by @ginacostag_ endorses HD Model 3.1 as a production-ready upgrade, highlighting markedly improved detail, geometry refinement, and texture clarity.

Tripo's single-image multi-view generation recreates occluded surfaces (rear, lateral, and underside) with surprisingly high fidelity, per @ginacostag_ (2026-06-15).

Tripo's 1st Anniversary Special offers 500 bonus credits when signing up with invite code Iucdiv (link provided).

Gina Costag highlights Factory’s focus on the entire software lifecycle, not just code, pull requests, or tests.

GBrain is an MIT-licensed open-source retrieval layer for AI agents, described as behaving "like Postgres" and acting as a just-in-time "3 book librarian" to supply contextual information.

GBrain is available as a hosted demo at gbrain.io, which Garry Tan (/@garrytan) recommended on X/Twitter on 2026-06-15 as the fastest way to try the product.

Paula (@paularambles) announces the launch of Souperintelligence (souperintelligence.com) on 2026-06-15, warning that highly capable AI systems could appear within a year — even "possibly by dinner" — and pledging to focus the project on keeping those systems aligned with human taste.

Factory announced Factory 2.0 on June 15, 2026, framing the update as a platform evolution “from coding agents to software factories.” The announcement (shared with a promotional video) was reshared by @EXM7777, who reacted enthusiastically (“this is lowkey incredible”), directing readers to the original nitter link for the full post and video.

Higgsfield released a plugin for Photoshop (announced 2026-06-15) that embeds its AI into the app: it converts live sketches into finished artwork, decomposes flat images into editable, fully reconstructed layers (recovering hidden pixels where objects overlap), and restyles photos via presets or prompts—presented in a short demo video and touted as a major productivity win for designers.

Diptish09 posted a brief promotional tweet on June 15, 2026, urging followers to "Try now" and linking to https://artguru.ai/diptisharma/.

OrcaRouter announced on June 15, 2026 that after OpenRouter's Fusion release (~48 hours earlier) they rebuilt the desirable parts of Fable 5 into OrcaRouter.ai DSL: a YAML-configurable graph that fans one prompt to multiple panels, runs a judge plus synthesizer, and returns a single consolidated answer—positioned as cheaper, open, and under user control, with stronger coding reliability than Fusion.

Author @Diptish09 tested WeArtguru on an old, damaged family photo and found the tool restored missing details, improved overall image quality, and upscaled the result to 8K.

Factory AI announced 'Factory 2.0' on 2026-06-15, framing the evolution as a move 'from coding agents to software factories.' The brief video post from @FactoryAI signals a strategic shift from standalone coding assistants toward an integrated platform for end-to-end software production and automation.

David Ondrej warns that "the gap is widening fast," arguing on 2026-06-15 that people who do not run the model Le Chaton Fat locally are at a growing disadvantage.

Hermes Agent gained a claimed tenfold increase in capability after integrating a single MCP, according to @DavidOndrej1 (2026-06-15).

Author @DataChaz shared a brief Twitter/X post on 2026-06-15 directing readers to the Hugging Face model page for yuxinlu1/gemm (https://huggingface.co/yuxinlu1/gemm) and provided no further commentary, technical details, or stated priority in the message.

DataChaz announces that @tryselects is live and accessible at cutback.video/selects (posted 2026-06-15), urging followers to try it now.

@DataChaz posted a bare GitHub repo link (github.com/DeusData/codebase…) on 2026-06-15.

Vercel's AI SDK adds portable agent harnesses (Claude Code, Codex, Pi) to decouple agents from specific providers while preserving performance advantages tied to certain harnesses.

Tom Kim announces a ground-up rebuild of Premiere Pro tailored for AI agents, claiming it interprets footage and produces edits meeting professional standards.

Charlie Hebdo pushes a direct call to subscribe to its daily newsletter (twp.ai/4hrVQk) in a 2026-06-15 post, listing what subscribers will receive: front pages, cartoons, investigative pieces, portraits, on‑the‑ground reports, behind‑the‑scenes material and a weekly podcast, framed as a catch‑all for the paper's core content.

Chamath and 8090 emphasize that Control, Consistency, Governance, and Auditability will become critical for software.

Unsloth AI's Kimi K2.7 Code has been shrunk from 1T to a 325 GB Dynamic 2-bit build (important layers upcasted) that runs at over 40 tokens/sec on ~330 GB RAM/VRAM machines, with a full-precision option requiring ~610 GB.

Lovable's new toolbar, announced on X/Twitter on 2026-06-15, introduces four switchable modes — annotate, select elements, edit text, and add comments — to streamline in‑app editing and collaboration.

Author @anthonyriera posted a brief attribution on Twitter/X on 2026-06-15 stating that the account or project @DataFast_ is by @marclou.

Anthony Riera (@anthonyriera) touts rankhog.com’s intentionally unique design to deter or shame imitators and says the site’s distinctiveness protects his work.

The post by @alex_prompter advertises a paid Claude skills bundle that claims to turn Claude into over 20 specialist agents for marketing and business tasks, framing the product as 'real expertise, not just prompts' and driving readers to linktr.ee/alex_prompter for purchase or details.

The post promotes a 'Claude skills bundle' from @alex_prompter that converts Anthropic's Claude into more than 20 specialized marketing and business experts.

Kran Diash (posted 2026-06-15) announced Sonic-3.5 for text-to-speech and Ink-2 for speech-to-text, claiming they are the #1 streaming models for speaking and listening respectively.

Higgsfield’s new Photoshop plugin brings real-time sketch-to-image conversion, automatic decomposition of flat JPGs into editable layers (including inferred hidden content), and one-click restyling via presets or prompts.

Leland is being reintroduced by John Koelliker on 2026-06-15 as a product built for a labor market where careers change every few years.

The tweet from @AIFrontliner on 2026-06-15 directs readers to play.cartesia.ai, inviting them to test voices on Cartesia's demo site.

Sonic 3.5 is presented as a top-ranked TTS model (Artificial Analysis #1) supporting 42+ languages, with a fast 82 ms time-to-first-audio and roughly 70% preference in blind tests.

Adele Bloch launched a social-and-dating tips series on 2026-06-15, positioning the posts as a “dating manifesto” for anyone needing a reminder of their worth in relationships.

Author @0xSero (2026-06-15) praises the Grok build harness for use with local models, calling it phenomenal and highlighting its smoothness, slick UI/UX, and high speed.

Github repository by @emollick (github.com/emollick/superlum…) was uploaded on 2026-06-15 via Claude 4.8 Opus.

User @tynan praises the newly released Pebble watch (post dated 2026-06-15), calling it “unbelievable” and noting that within two days they built their ideal watchface.

Author @paulg (2026-06-15) recommends using DuckDuckGo for image searches, claiming its results are noticeably better.

The post by @lazukars endorses Gauth Atlas’s interactive history feature, pointing to a 2026-05-18 Gauth_AI tweet that invites users to “explore history instead of just reading it.” The feature visualizes Zhuge Liang’s life as an interactive landscape, emphasizing his roles as strategist and inventor and linking to gauthmath.com/atlas for users to create their own journey.

Charly Wargnier (@DataChaz) posted on 2026-06-04 promoting McDonald's GrimaceBot as a free alternative to paid models like Claude, claiming it will respond to any question beyond brand-related topics.

Poke (@interaction) is promoted by @lazukars (2026-06-05) as newly approved by Apple to text on Apple Messages, presented as the "first and only AI agent" on the platform.

Antioch Agent is a browser-based simulator that purports to recreate the full physical AI stack in a closed agentic loop, allowing engineers to onboard robots, construct high-fidelity scenes, and validate autonomy end-to-end.

@kyleanthony posted on 2026-06-06 announcing limited remaining stock and urging followers to buy by visiting his online store at store.brasshands.com/.

New Mexico production data is now available in the Wells Intelligence app and Clearfork Wells API, providing oil, gas, and produced-water histories from 1993 onward.

@kyle_e_walker is issuing a final reminder that May 22, 2026 is the last day to purchase his Census/mapping workshop catalog on Gumroad before the store closes.

freestiler now integrates H3 to generate multi-layer hexagonal vector tiles from point data.

Wells Intelligence, promoted by @kyle_e_walker on 2026-06-07, positions itself as a dual-purpose tool: a fast, interactive map for humans to explore TX/NM production data and a well API that can be handed to software agents for automated analysis.

The post promotes 'Coreweave alpha' attributed to @KookCapitalLLC and @HYPEconomist via TheKookLetter, a paid newsletter.

Tweet by @rtrdednigga (2026-06-14) endorses Kookletter, stating it comprehensively covers all finance sectors, delivers “high level alpha,” and hosts the author’s favorite Telegram group—described as the “best tg I’ve ever been in.” The post is a concise personal recommendation rather than an analytic review.

@KookCapitalLLC promotes a paid subscription called @TheKookLetter, using the phrase "non stop alpha" as the primary value claim and directing readers to whop.com/kookletter.

Multigres v0.1 (alpha), announced by @kiwicopple on 2026-06-04 as part of a fundraising round, is an open-source, self-hostable platform described as an “operating system” for PostgreSQL.

Kimi_Moonshot demonstrates a workflow that performs large-scale searches across multiple platforms and automatically fills the aggregated results into a spreadsheet.

K2.6's multimodal capability is presented as allowing an agent to open a website, navigate through it, and replicate it.

Kimi_Moonshot posted a short video (2026-05-14) showcasing an AI agent that observes a user’s daily workflows and turns those routines into reusable skills.

Kimi_Moonshot demonstrates an AI agent that, via chat, constructs a complete Google Form and uses a browser to perform the typing and form-building steps automatically.

Kimi announced a WebBridge feature that connects its platform to several code-oriented models and tooling—Kimi Code CLI, Claude Code, Cursor, Codex, Hermes, and others—enabling users to try it at kimi.com/features/webbridge or install it from the Chrome Web Store.

Kimi Code K2.6 is an open-source coding agent offering a one-line CLI install and immediate startup; it adds video-as-context features (reference-to-LUT, long-video-to-short, screen-recording-to-code), plugin support for finance and academic sources, ACP protocol compatibility, IDE integrations (JetBrains, Zed), and developer hooks for custom tools and workflows.