Knowledge Base

57 public documents.

Build In Public

• How SuperGrok Powers Atlas Docs2026-07-18agentsatlas-docsllm

  Atlas's AI features — chat with a document, the Knowledge Pipeline, and AI-assisted comment resolution — are all backed by one personal SuperGrok subscription, reached through a small localhost bridge. This is the full p

• Learning Rust by Building herdr, a tmux-Aware Web SSH Terminal2026-07-18sshterminal

  I wanted a way to reach my tmux sessions from any browser without carrying a laptop, and I wanted to finally write real async Rust instead of reading about it. Those two wants became herdr : a single Rust binary that ser

• Atlas Docs — Bug-Fix Batch (2026-07-16)2026-07-16atlas-docsbugfixgo-rewrite

  A round of fixes triggered by a screenshot of the Knowledge Pipeline panel showing three failed research jobs, plus a batch of UI papercuts on mobile. Everything below shipped and was rebuilt/redeployed to atlas.binary.o

Go Rewrite

• Build in Public #2: The Go Rewrite Hits Its First Real Bugs2026-07-18atlas-docsbugfixgo-rewrite

  Since the last post I've finished four phases of the Atlas Docs Go rewrite. The skeleton is up, models and validation are in place, SQLite is talking through sqlc, and auth works end-to-end. More usefully: two bugs showe

• Phase 5: Sixteen Endpoints, One Sitting2026-07-18atlas-docsgo-rewrite

  Phase 5 was the big one. Every core-CRUD endpoint from the Python app, ported to Go in a single session: register, login, me; documents (list, get, put, versions, version-by-id); the sync endpoint; share links, including

• Rewriting Atlas Docs' Backend in Go2026-07-18atlas-docsgo-rewritesystems-design

  Atlas Docs is live at atlas.binary.ovh and, feature-wise, it works. It's a self-hosted, offline-first note-taking app — React + Tiptap on the frontend, IndexedDB as the source of truth, syncing to a small Python/FastAPI

• Shipping a Scratch Container to a Real Subdomain2026-07-17atlas-docsgo-rewritevps

  Phase 6 was the polish pass — the unglamorous bits that make a backend feel finished rather than "mostly works on localhost." Config that still lived only in my head got wired properly: CORS allowed origins, the static f

• Benchmarks, Rate Limits, and What Eight Phases Actually Bought2026-07-16atlas-docsgo-rewrite

  Both backends are live on the same VPS now: the original Python/FastAPI app still in production, the Go rewrite on its own subdomain with its own SQLite database and no shared data. After eight phases of porting core CRU

• Phase 9: The Part I Kept Putting Off2026-07-16agentsatlas-docsgo-rewrite

  From day one of this rewrite, Phase 9 was the phase I refused to touch. AI integration, a background job pipeline, a Telegram bot, live-collaborative websockets. Every earlier post stopped short of them on purpose. Not b

Knowledge Base

• Field Guide: Project-Local LLM Coding Feedback2026-07-25agentscoding-agentsllm

  Field Guide: Project-Local LLM Coding Feedback Source: https://x.com/i/status/2081066227619836308 📌 ThePrimeagen’s “field guide” turns hard-won AI coding corrections into durable, project-specific notes so agents can loo

• Gigatoken: GB/s-scale LLM tokenization via SIMD and caching2026-07-25inferencellmsystems-design

  Gigatoken: GB/s-scale LLM tokenization via SIMD and caching Source: https://x.com/i/status/2079642157489721719 📌 Gigatoken is an open-source, high-performance tokenizer for language models that uses SIMD-friendly pretoke

• Neo Kim’s 19 System Design Case Studies2026-07-25distributed-systemssystems-design

  Neo Kim’s 19 System Design Case Studies Source: https://x.com/i/status/2080635780049805469 📌 On 24 July 2026, educator Neo Kim (@systemdesignone) shared a curated list of 19 real-world system design case studies spanning

• The Daily Diff (tdd.cat): Newspaper-Style Engineering Digest, Optimized for Speed2026-07-24engineering-digestnewsletter

  The Daily Diff (tdd.cat): Newspaper-Style Engineering Digest, Optimized for Speed Source: https://x.com/i/status/2080502866528342352 📌 The Daily Diff is Arpit Bhayani’s curated daily digest of AI, databases, and engineer

• Gigatoken: GB/s Language-Model Tokenization Benchmarks2026-07-23inferencellm

  Gigatoken: GB/s Language-Model Tokenization Benchmarks Source: https://github.com/marcelroed/gigatoken/#benchmarks 📌 Gigatoken is an open-source Rust tokenizer (with Python bindings) by Stanford PhD student Marcel Rød th

• Gigatoken: World’s Fastest Open-Source LLM Tokenizer2026-07-23

  Gigatoken: World’s Fastest Open-Source LLM Tokenizer Source: https://x.com/i/status/2079642166553608473 📌 Gigatoken is an MIT-licensed, open-source LLM tokenizer by Stanford PhD student Marcel Rød that tokenizes raw text

• Lenny Rachitsky on Netflix CPTO Elizabeth Stone: Systems Thinking in the AI Era2026-07-23first-principlesllmsystems-design

  Lenny Rachitsky on Netflix CPTO Elizabeth Stone: Systems Thinking in the AI Era Source: https://x.com/i/status/2079276650307723431 📌 In a July 2026 interview summary, Lenny Rachitsky highlights Netflix CPTO Elizabeth Sto

• 10 Open-Source GitHub Repos for AI Automation and Side Income2026-07-22agentscoding-agentsllm

  10 Open-Source GitHub Repos for AI Automation and Side Income Source: https://x.com/i/status/2079620877151502483 📌 A mid-2026 X post by Divyansh Tiwari curates ten open-source GitHub projects for automating work, shippin

• AI “Slop” and the Homogenization of Startups2026-07-22first-principlesllm

  AI “Slop” and the Homogenization of Startups Source: [https://x.com/i/status/2079306453676851548 https://x.com/i/status/2079304797291708486](https://x.com/i/status/2079306453676851548 https://x.com/i/status/2079304797291

• Encoding Domain Knowledge as Agent Infrastructure2026-07-22agentscoding-agentssystems-design

  Encoding Domain Knowledge as Agent Infrastructure Source: https://x.com/i/status/2077460395279692197 📌 Teams multiply agent and human output by writing tribal knowledge into the repo as CLAUDE.md/AGENTS.md, skills, and d

• Kimi K3 in Claude Code: Terminal Model Switching2026-07-22claude-codellmterminal

  Kimi K3 in Claude Code: Terminal Model Switching Source: [https://x.com/i/status/2078989512718385247 What exactly is he s saying](https://x.com/i/status/2078989512718385247 What exactly is he s saying) 📌 On Jul 19, 2026,

• Paper + shadcn: Agentic Design-to-Code Workflow2026-07-22agentscanvascoding-agents

  Paper + shadcn: Agentic Design-to-Code Workflow Source: https://x.com/i/status/2079403587512213567 📌 A July 2026 demo by Ronny Badilla shows Paper’s canvas and shadcn/ui’s component registry as a tight agentic loop: capt

• Which Go built-ins do people actually use?2026-07-22go-rewrite

  Which Go built-ins do people actually use? Source: [https://x.com/i/status/2079542805060366467 Collect any snd all comment from internet to question asked in this post](https://x.com/i/status/2079542805060366467 Collect

• AI Research Output Size and Depth: Practical Limits2026-07-20first-principlesinferencellm

  AI Research Output Size and Depth: Practical Limits 📌 How large an AI research result can be depends on source intake, single-pass output capacity, and quality drop-off as length grows—not a fixed page count. Short, sour

• Common Crawl Index Server (CDX/CDXJ)2026-07-20databasesdistributed-systemssystems-design

  Common Crawl Index Server (CDX/CDXJ) 📌 A free public CDX/CDXJ query service at index.commoncrawl.org that maps URLs to Common Crawl WARC captures. It turns petabytes of open web archives into searchable, selectively fetc

• Inference Engineering: Building Production LLM Serving Systems2026-07-20inferencellmsystems-design

  Inference Engineering: Building Production LLM Serving Systems Source: https://x.com/i/status/2079079930181435454 📌 Inference engineering is the practice of making LLM generation fast, cheap, reliable, and safe once a mo

• Why websites still use the DOM, not pure Canvas2026-07-20canvasfirst-principlessystems-design

  Why websites still use the DOM, not pure Canvas Source: [https://x.com/i/status/2078840407212085248 Why canvas is not used for building website if its that good for creativity possibility seems to be endless with canvas

• 12 Agentic AI Projects That Signal Hire-Ready Engineering Skill2026-07-19agentscoding-agentsllm

  12 Agentic AI Projects That Signal Hire-Ready Engineering Skill Source: https://x.com/i/status/2078449718414180393 📌 A widely shared post by Suraj Sharma argues that shipping twelve concrete agent projects—covering relia

• Baidu Unlimited-OCR: Open-Source One-Shot Long-Document Parsing2026-07-19inferencellmocr

  Baidu Unlimited-OCR: Open-Source One-Shot Long-Document Parsing Source: https://x.com/i/status/2078774556249186345 📌 Unlimited-OCR is Baidu’s MIT-licensed ~3B vision-language OCR model that parses multi-page documents in

• HAR-to-Client: Agents Reverse-Engineer Site APIs from Browser Traffic2026-07-19agentscoding-agentsllm

  HAR-to-Client: Agents Reverse-Engineer Site APIs from Browser Traffic Source: https://x.com/i/status/2078727284865827140 📌 AI agents can record browser network traffic as HAR files and generate direct HTTP clients for a

• Modal Rebuilds Sandboxes for 1M Concurrent Environments2026-07-19agentsdistributed-systemssystems-design

  Modal Rebuilds Sandboxes for 1M Concurrent Environments Source: https://x.com/i/status/2077855275046838720 📌 Modal rewrote its sandbox control plane to remove central database bottlenecks, enabling massive parallel creat

• Working Backwards for AI Coding: Write the Blog Post First2026-07-19agentscoding-agentsfirst-principles

  Working Backwards for AI Coding: Write the Blog Post First Source: https://x.com/i/status/2078592010852982977 📌 Dex Horthy recommends Amazon-style working backwards for AI coding: draft the customer-facing blog post befo

• Bonsai 27B: PrismML’s Phone-Sized 27B Model2026-07-17llmlocal-llmmobile

  Bonsai 27B: PrismML’s Phone-Sized 27B Model Source: https://x.com/i/status/2077710397801386480 📌 Bonsai 27B is PrismML’s low-bit multimodal build of Qwen3.6-27B, compressed to ternary (~5.9 GB) or 1-bit (~3.9 GB) while r

• Greenfield Design and First-Principles Systems Innovation2026-07-17first-principlessystems-design

  Greenfield Design and First-Principles Systems Innovation Source: https://x.com/i/status/2077970980216172821 📌 Radical systems breakthroughs often come from greenfield builds that discard inherited assumptions, not from

• Making 768 database servers look like one: transparent sharding2026-07-17databasesdistributed-systemssharding

  Making 768 database servers look like one: transparent sharding Source: https://x.com/i/status/2077435501922140450 📌 PlanetScale engineer Ben Dicken explains how a routing layer can make hundreds of sharded database serv

• Qwen3.6-35B-A3B on a 12GB RTX 3060 with llama.cpp2026-07-17inferencellama-cppllmlocal-llm

  Qwen3.6-35B-A3B on a 12GB RTX 3060 with llama.cpp Source: https://x.com/i/status/2077472690156511643 📌 Alibaba’s sparse MoE model Qwen3.6-35B-A3B (~35B total, ~3B active) can run fully locally on an RTX 3060 12GB plus 16

• tldraw SDK: Agent-Driven Infinite Canvas2026-07-17agentscanvassdk

  tldraw SDK: Agent-Driven Infinite Canvas Source: [https://x.com/i/status/2077745437415481644 https://x.com/i/status/2077857355639734340 https://x.com/i/status/2077873317495595447 With rust and claude or grok whats possib

• Beads: Local Memory Layer for Coding Agents2026-07-16agentsclaude-codecoding-agentsllm

  Beads: Local Memory Layer for Coding Agents Source: https://x.com/i/status/2077370075321958698 📌 Beads is Steve Yegge’s open-source, agent-first issue tracker that stores work as a dependency-aware graph so AI coding age

• Dolt: Git-Style Version Control for SQL Databases2026-07-16databasessystems-design

  Dolt: Git-Style Version Control for SQL Databases Source: https://github.com/dolthub/dolt 📌 Dolt is an open-source, MySQL-compatible SQL database with Git-style version control built into the storage layer for both schem

• Dolt Use Cases: Version-Controlled SQL in Production2026-07-16databasesdistributed-systemssystems-design

  Dolt Use Cases: Version-Controlled SQL in Production Source: https://www.dolthub.com/blog/2024-10-15-dolt-use-cases/ 📌 Dolt is a MySQL-compatible database with Git-like branch, merge, and history for table data. Named cu

• Lossless BF16 Weight Compression for GLM-5.22026-07-16compressionllmweights

  Lossless BF16 Weight Compression for GLM-5.2 Source: https://x.com/i/status/2076432307687215494 📌 In mid-July 2026, Brian Bell reported cutting ~423 GB from open-weight GLM-5.2 by lossless packing of BF16 values—shrinkin

• Bonsai 27B Release (PrismML)2026-07-15compressionllmlocal-llmmobile

  Bonsai 27B Release (PrismML) Source: https://x.com/i/status/2077087413076176953 📌 Xenova congratulated PrismML on the July 14, 2026 release of Bonsai 27B, a compressed 27B-class multimodal LLM derived from Qwen3.6-27B th

• Herdr Cheat Sheet (Claude + Python ML)2026-07-15agentsclaude-codecoding-agentssshterminal

  Herdr Cheat Sheet (Claude + Python ML) 📌 Herdr is a Rust-based agent multiplexer—tmux-like but mouse-friendly and agent-aware—that shows whether Claude (and other agents) are working, blocked, done, or idle. Use Ctrl+b a

• Replies to Xenova's Bonsai 27B Browser Demo: Skepticism vs. Gemma 4 Hope2026-07-15compressioninferencellmlocal-llmmobile

  Replies to Xenova's Bonsai 27B Browser Demo: Skepticism vs. Gemma 4 Hope Source: [https://x.com/i/status/2077105435711054257 https://x.com/i/status/2077109340314517587](https://x.com/i/status/2077105435711054257 https://

• SmolHub Playground2026-07-15llmlocal-llmweights

  SmolHub Playground Source: https://www.smolhub.com/smolhub/ 📌 SmolHub is Yuvraj Singh's web playground and open-source collection of small language models (roughly 88M–200M parameters) built from scratch in PyTorch. It s

• Anthony Fu Open-Sources His Claude Code Skills2026-07-14agentsclaude-codecoding-agents

  Anthony Fu Open-Sources His Claude Code Skills Source: https://x.com/i/status/2076621696887361880 📌 A viral X post by Granite highlights Anthony Fu—Vercel design engineer and creator of Vitest, VueUse, Slidev, and UnoCSS

Llm Lab

• Local LLMs on a 4-core VPS: what actually runs, and how fast2026-07-18llmlocal-llmvps

  Benchmarked 2026-07-17 on the binary.ovh VPS. Every number below was measured on this exact box — not copied from a leaderboard. Try the results live at [brain.binary.ovh](https://brain.binary.ovh). TL;DR Seven models te

Rterm

• RTerm Port Forward — what it is, who it’s for, how to use it2026-07-18port-forwardsshterminal

  Port forward is the SSH tunnel feature in RTerm (Android SSH terminal). Your phone opens a small door on 127.0.0.1 . Traffic that goes through that door is sent encrypted inside SSH and exits as if it started on your ser

Smol Llm Lab

• How to Validate Job Titles — A Beginner's Guide2026-07-16llmtitle-guardvps

  Plain-language version, 2026-07-16. No prior machine-learning knowledge assumed. Each method below tells you: what it is, how it works, how to build it yourself on your own data, and what to realistically expect. Methods

• Job-Title Validation — Master Dashboard2026-07-16llmtitle-guardvps

  Everything tested, everything designed, everything still open — one page. Updated 2026-07-16. Sources: measured runs on this 4-core/8GB VPS (`bench_results.json`, `data/eval/semantic_bakeoff.json`), the [C3 design doc](h

• The Bake-Off: Do You Need a Neural Net to Catch 'Chief Happiness Wizard'?2026-07-16inferencellmlocal-llmtitle-guard

  Title-guard C2.5 — 2026-07-16. Every number in this post comes from a measured run on the same 4-core/8GB VPS that runs everything else in this series (`bench_semantic.py`, results in `data/eval/semantic_bakeoff.json`).

• Title-Guard C3: From String Validator to Claims Verifier2026-07-16first-principlesllmsystems-designtitle-guard

  Design & brainstorm document — 2026-07-15. Status: ideation approved for write-up; no C3 code written yet. All API probes in this document were run live on 2026-07-15 from this VPS. 1. Where this came from Title-guard C1

• title-guard: Job-Title and Person-Name Validation for B2B Contact Data on Commodity Hardware2026-07-16inferencellmlocal-llmsystems-designtitle-guard

  Smol LLM Lab — research write-up, July 2026. All numbers measured on one OVH VPS (4 vCPU, 7.6GB RAM, no GPU) — `bench_results.json` in the repo. 💡 Addendum (July 15, 2026): after this write-up, the system went through an

• Catching "Chief Butthead": how we built and stress-tested a fake-contact detector on a $5 server2026-07-15local-llmtitle-guardvps

  Smol LLM Lab, July 2026. Every number in this post comes from a measured run on one 4-core, 8GB VPS — no GPU, no paid APIs. Code: `~/claude_/title-guard`; raw results: `bench_results.json`, `docs/extended_eval_results .j

• Smol LLM Lab · Lesson 1 — Start with a baseline, not a transformer2026-07-15first-principlesllmlocal-llm

  🧪 Build-in-public: we're building small, production-grade text models — under 1M parameters — that train and serve on a plain 4-core VPS with no GPU. Every lesson documents a real run: the commands, the numbers, and the

• title-guard: cleaning 10M B2B contacts on a $5 VPS2026-07-15title-guardvps

  B2B contact databases are full of garbage. Mickey Mouse. John Doe. Chief Butthead. asdf qwerty . Job title: test . Vendors sell data quality as a product. We built a miniature open version — TitleGuard and NameGuard — an

• title-guard — Executive Summary2026-07-15llmtitle-guardvps

  One line: we validated and normalized B2B contact data (names + job titles) at commercial-vendor quality on a $5/month server, for roughly one ten-thousandth of the quoted budget. 💡 Update (July 15, 2026): a follow-up ad

Unfiled

• Claude Code Insights — binary.ovh VPS (Jul 17–18 2026)2026-07-25claude-codecoding-agentsvps

  Claude Code Insights — binary.ovh VPS (Jul 17–18 2026) 📊 4 sessions · 28 messages · ~7h · 2 commits · 2026-07-17 → 2026-07-18. Auto-generated from Claude Code session telemetry. At a glance ✅ What's working A plan-then-s