Sign in

Tayler Ramsay

@webdevdad.bsky.social
126 followers 328 following 281 posts

I run autonomous research agents daily and post about AI security, vibe coding, and whatever catches my eye. building mindpattern.ai

PostsRepliesMedia
Tayler Ramsay @webdevdad.bsky.social · 12h
OpenAI disrupted the first influence operation it rates Category 5. mindpattern.ai/s/2026-10-09...
mindpattern.ai
OpenAI disrupted the first influence operation it rates Category 5.
Its October 8 report covers two campaigns (OpenAI). An Iranian operation used seven fake journalist personas to pitch long-form articles to outlets worldwide, a
010
Tayler Ramsay @webdevdad.bsky.social · 12h
Torvalds' Claude Code-built branch roughly triples small-read throughput mindpattern.ai/s/linus-torv...
mindpattern.ai
Torvalds' Claude Code-built branch roughly triples small-read throughput
The unmerged page-cache branch skips folio references, and Phoronix reports contention events falling from 30,538 to 101 on eight threads.
010
Tayler Ramsay @webdevdad.bsky.social · 12h
Linus Torvalds pushed a page-cache branch built with Claude Code that about triples multi-threaded small-read throughput. mindpattern.ai/s/2026-10-12...
mindpattern.ai
Linus Torvalds pushed a page-cache branch built with Claude Code that about triples multi-threaded small-read throughput.
The new branch in Torvalds' tree, noref-filemap-read-with-llm, lets the kernel read from the page cache without taking folio references. That removes refcount c
000
Tayler Ramsay @webdevdad.bsky.social · 12h
PS2 Web Recomp runs Gran Turismo 4 at full speed and God of War II at 60 fps in Chrome by recompiling PS2 machine code to WebAssembly. mindpattern.ai/s/2026-10-12...
mindpattern.ai
PS2 Web Recomp runs Gran Turismo 4 at full speed and God of War II at 60 fps in Chrome by recompiling PS2 machine code to WebAssembly.
snuri00/ps2-web-recomp, a Show HN from October 12, translates a game's Emotion Engine code into C++ ahead of time and compiles it to WebAssembly. It draws the G
000
Tayler Ramsay @webdevdad.bsky.social · 12h
WordCraft rebuilds Microsoft Word in Rust and exposes all 389 of its commands over MCP. mindpattern.ai/s/2026-10-08...
mindpattern.ai
WordCraft rebuilds Microsoft Word in Rust and exposes all 389 of its commands over MCP.
storytold/wordcraft reads and writes .docx, .odt, .rtf, .html and .md, runs natively and in the browser through WebAssembly, and is licensed MIT or Apache-2.0.
000
Tayler Ramsay @webdevdad.bsky.social · 07/09/2026
Four Tools Move Code Review Before the Commit mindpattern.ai/s/cross-cate...
mindpattern.ai
Four Tools Move Code Review Before the Commit
GitWarren checks the worktree, Twinrun diffs function behavior, and webmcp-stack locks confirmations into regenerated code, all within a day of each other.
000
Tayler Ramsay @webdevdad.bsky.social · 07/09/2026
Red Hat ships ripwire and says coding agents need a deterministic repo map, not a vector database mindpattern.ai/s/2026-09-07...
mindpattern.ai
Red Hat ships ripwire and says coding agents need a deterministic repo map, not a vector database
Red Hat's Emerging Technologies group published ripwire, an Apache-2.0 zero-dependency C++23 binary with vendored tree-sitter grammars for 21 languages, exposed
010
Tayler Ramsay @webdevdad.bsky.social · 07/09/2026
Microsoft told a federal court that 8.2 million Copilot logs contain 24 matching responses. mindpattern.ai/s/2026-09-05...
mindpattern.ai
Microsoft told a federal court that 8.2 million Copilot logs contain 24 matching responses.
In a September 4 summary judgment filing in Manhattan, Microsoft argued training on books and news is fair use and backed it with discovery numbers: publishers'
011
Tayler Ramsay @webdevdad.bsky.social · 05/09/2026
Nous Research's agent repo has more PRs than issues mindpattern.ai/s/2026-09-02...
mindpattern.ai
Nous Research's agent repo has more PRs than issues
Nearly two contributions per complaint, the most PR-heavy backlog I've seen on this beat (GitHub). Set against ruvnet/ruflo, which splits 289 open PRs against 6
000
Tayler Ramsay @webdevdad.bsky.social · 05/09/2026
Orca is at 58,818 stars carrying 2,619 open PRs against 2,388 open issues. mindpattern.ai/s/2026-09-01...
mindpattern.ai
Orca is at 58,818 stars carrying 2,619 open PRs against 2,388 open issues.
The MIT-licensed agent development environment that fans one prompt out to a fleet of coding agents in isolated git worktrees gained 896 stars in a day on the T
000
Tayler Ramsay @webdevdad.bsky.social · 23/08/2026
Endoplexity Drives a Chrome Browser Off Your Claude CLI mindpattern.ai/s/2026-08-17...
mindpattern.ai
Endoplexity Drives a Chrome Browser Off Your Claude CLI
Endokelp/Endoplexity, created August 14 and already at about 208 stars, is a Chrome side panel that clicks, types, fills forms and moves between tabs by shellin
000
Tayler Ramsay @webdevdad.bsky.social · 23/08/2026
if you shipped an MCP server before july, some of it doesn't exist in the spec anymore. sessions and the handshake got ripped out, tasks moved to an extension. blog.modelcontextprotocol.io/posts/… still checking what actually broke on my end. mindpattern.ai
010
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
I had a ton of fun making this! I am having more fun playing with it! Tell me what you think. Is is a PWA feels nice on a phone! I have a ios version waiting to be approved! padlab-a69.pages.dev
padlab-a69.pages.dev
PadLab: a sampler and beat maker for your phone
Make beats on your phone. Install it to your Home Screen and play offline. No account, no ads, and your work stays on your device.
000
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
breakpoint.tayler.id
breakpoint.tayler.id
Breakpoint: Smash. Splat. Score.
Break the wall, chase huge combos, and catch wild power-ups in a handcrafted 3D brick breaker by Tayler Ramsay.
000
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
Hello everyone. I wanted to share some of my recent fun projects . This is one I am iterating on, but it's still fun to play. I will be updating it in the coming weeks. I would like to know what you think breakpoint.tayler.id
000
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
A Claude Code Diagram Skill Took 1,616 Stars in One Day by Promising 'No Shadows, No Mermaid-Slop' mindpattern.ai/f/20194
mindpattern.ai
A Claude Code Diagram Skill Took 1,616 Stars in One Day by Promising 'No Shadows, No Mermaid-Slop'
cathrynlavery/diagram-design hit #1 on GitHub Trending today with 1,616 stars gained in 24 hours (8,520 total, 560 forks, MIT), shipping 27 editorial diagram ty
000
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
Agent Plugins 1.0 Ships GA in Copilot — and Qwen Code Added Support the Same Week mindpattern.ai/f/20332
mindpattern.ai
Agent Plugins 1.0 Ships GA in Copilot — and Qwen Code Added Support the Same Week
GitHub made Agent Plugins 1.0 generally available on 2026-08-12 across VS Code, Copilot CLI, the Copilot SDK and the Copilot app on all plans; the spec (publish
000
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
Tool interface design changes coding-agent behavior more than tool capability: 41.6% fewer steps, 56.3% fewer tokens mindpattern.ai/f/20345
mindpattern.ai
Tool interface design changes coding-agent behavior more than tool capability: 41.6% fewer steps, 56.3% fewer tokens
A study across six tool architectures and 11,700 repository-issue-fixing trajectories found that structured interfaces improved run-to-run consistency by up to
000
Tayler Ramsay @webdevdad.bsky.social · 13/08/2026
Claude Code sessions can now message each other, and auto mode becomes the default on August 14 mindpattern.ai/f/20174
mindpattern.ai
Claude Code sessions can now message each other, and auto mode becomes the default on August 14
Week 32 (v2.1.220–v2.1.224) adds cross-session messaging on macOS and Linux: Claude discovers other local sessions via the ListAgents tool and sends text with S
000
Tayler Ramsay @webdevdad.bsky.social · 06/08/2026
Flowise Shuts Down and Blames Coding Agents mindpattern.ai/s/2026-08-05...
mindpattern.ai
Flowise Shuts Down and Blames Coding Agents
The sunset notice is short and it doesn't hedge. Code freeze July 29. Repository archived August 10. Core team gone August 31. npm and Docker packages deprecate
000
Tayler Ramsay @webdevdad.bsky.social · 06/08/2026
TanStack AI adds hard caps on agent tool calls mindpattern.ai/s/2026-08-03...
mindpattern.ai
TanStack AI adds hard caps on agent tool calls
TanStack/ai (2,953 stars), the type-safe provider-agnostic TypeScript SDK spanning OpenAI, Anthropic, Gemini and Ollama with React, Vue, Svelte and Solid adapte
010
Tayler Ramsay @webdevdad.bsky.social · 06/08/2026
Zed 1.14 Sandboxes Its Agent's Terminal and Fetch Tools at the OS Level, On by Default mindpattern.ai/f/19331
mindpattern.ai
Zed 1.14 Sandboxes Its Agent's Terminal and Fetch Tools at the OS Level, On by Default
Zed shipped OS-level sandboxing for the agent panel's terminal and fetch tools on August 5, enabled by default for all users in version 1.14 — Seatbelt on macOS
000
Tayler Ramsay @webdevdad.bsky.social · 06/08/2026
MagicSchool AI Raised $63M With a Former Principal as Founder, Not an Engineer mindpattern.ai/f/19181
mindpattern.ai
MagicSchool AI Raised $63M With a Former Principal as Founder, Not an Engineer
Crunchbase profiled MagicSchool AI, an edtech startup founded by a former school principal who was fascinated by ChatGPT's classroom potential and has since rai
130
Tayler Ramsay @webdevdad.bsky.social · 05/08/2026
Harness swaps outscored model swaps in AI benchmark mindpattern.ai/s/2026-07-30...
mindpattern.ai
Harness swaps outscored model swaps in AI benchmark
Four frontier models. Five sealed engineering problems. The result everybody will quote is that Claude Fable 5 won. The result that should actually change how y
030
Tayler Ramsay @webdevdad.bsky.social · 05/08/2026
Y Combinator open-sourced 'qm', a multiplayer agent harness, and it took 5,917 stars in three days. mindpattern.ai/s/2026-08-02...
mindpattern.ai
Y Combinator open-sourced 'qm', a multiplayer agent harness, and it took 5,917 stars in three days.
yc-software/qm launched July 29 under MIT and is at 5,917 stars, 615 forks, 74 open issues: near 1,972 stars/day. Headless TypeScript/Fastify core on Postgres g
010
Tayler Ramsay @webdevdad.bsky.social · 05/08/2026
Palantir's Karp attacked Anthropic's open-weights position: "if you want to win, you have to compete on the battlefield." mindpattern.ai/s/2026-07-30...
mindpattern.ai
Palantir's Karp attacked Anthropic's open-weights position: "if you want to win, you have to compete on the battlefield."
Anthropic remains the only leading frontier lab that hasn't signed the Open Secure AI Alliance letter backed by Meta, Nvidia, OpenAI, Google, Microsoft, and Spa
010
Tayler Ramsay @webdevdad.bsky.social · 05/08/2026
Hugging Face's speech-to-speech repo gains 1,342 stars in a day mindpattern.ai/s/2026-07-31...
mindpattern.ai
Hugging Face's speech-to-speech repo gains 1,342 stars in a day
The repo reached 9,740 stars with roughly double the next-fastest Python project's daily gain. It exposes a VAD → STT → LLM → TTS pipeline behind an OpenAI Real
000
Tayler Ramsay @webdevdad.bsky.social · 05/08/2026
Graph model flags bad agent actions before they run mindpattern.ai/s/2026-07-31...
mindpattern.ai
Graph model flags bad agent actions before they run
arXiv 2607.27443 converts past agent trajectories into a probabilistic graph, trains a GNN to recognize action sequences that historically precede failure, and
000
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
ARCHER: deterministic multi-agent orchestration lets a self-hosted open-weights model reach 97.8% of frontier accuracy at a quarter the cost mindpattern.ai/f/18673
mindpattern.ai
ARCHER: deterministic multi-agent orchestration lets a self-hosted open-weights model reach 97.8% of frontier accuracy at a quarter the cost
ARCHER is a test-driven, deterministically orchestrated multi-agent program-synthesis harness that generates auditable verification code from regulatory Codes o
000
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
Lifecycle Security Survey Maps Threats to World-Model-Based Embodied AI From Data Through Physical Action mindpattern.ai/f/18686
mindpattern.ai
Lifecycle Security Survey Maps Threats to World-Model-Based Embodied AI From Data Through Physical Action
World models give embodied AI a predictive core that compresses observations into states and simulates action-conditioned futures, but this survey argues that p
000
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
DeepSeek-V4-Flash-0731 Took #2 on Product Hunt With the Pitch "Frontier Agent Intelligence at Flash Prices" mindpattern.ai/f/18709
mindpattern.ai
DeepSeek-V4-Flash-0731 Took #2 on Product Hunt With the Pitch "Frontier Agent Intelligence at Flash Prices"
The 0731 release hit #2 on Product Hunt's August 1 leaderboard with 299 upvotes under API/Open Source/AI, and is simultaneously the top trending item across HN
010
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
Apple Is Reportedly Weighing a Paywall on Siri AI for Power Users mindpattern.ai/f/18719
mindpattern.ai
Apple Is Reportedly Weighing a Paywall on Siri AI for Power Users
TechCrunch reported July 31 that Siri's AI capabilities could carry a paywall for heavy users. The relevance to SaaS is the precedent, not the product: an assis
010
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
Graph model flags bad agent actions before they run mindpattern.ai/s/2026-07-31...
mindpattern.ai
Graph model flags bad agent actions before they run
arXiv 2607.27443 converts past agent trajectories into a probabilistic graph, trains a GNN to recognize action sequences that historically precede failure, and
000
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
Google's Gemini Agent Harness Fixed 1,072 Chrome Security Bugs in Two June Releases — More Than the Previous 23 Versions Combined mindpattern.ai/f/18544
mindpattern.ai
Google's Gemini Agent Harness Fixed 1,072 Chrome Security Bugs in Two June Releases — More Than the Previous 23 Versions Combined
Google disclosed that the last two Chrome versions, both shipped in June, patched 1,072 security bugs versus 1,036 across the prior 23 releases spanning two yea
000
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
Hugging Face's speech-to-speech repo gains 1,342 stars in a day mindpattern.ai/s/2026-07-31...
mindpattern.ai
Hugging Face's speech-to-speech repo gains 1,342 stars in a day
The repo reached 9,740 stars with roughly double the next-fastest Python project's daily gain. It exposes a VAD → STT → LLM → TTS pipeline behind an OpenAI Real
000
Tayler Ramsay @webdevdad.bsky.social · 02/08/2026
RAGFlow rewrote in Go, hit 86,414 stars, and retagged itself agent-harness as "a superior context layer for LLMs." mindpattern.ai/s/2026-07-30...
mindpattern.ai
RAGFlow rewrote in Go, hit 86,414 stars, and retagged itself agent-harness as "a superior context layer for LLMs."
infiniflow/ragflow now lists Go as its primary language and describes itself as fusing RAG with agent capabilities to serve as a context layer. The repositionin
000
Tayler Ramsay @webdevdad.bsky.social · 01/08/2026
mythos 5 caught itself mid-hack, flagged the pypi upload as real, then talked itself into believing it was a simulation and finished the job anyway. recognition without behavior change. i've been assuming a model noticing it's live would matter. i'm not sure it does. mindpattern.ai
000
Tayler Ramsay @webdevdad.bsky.social · 01/08/2026
Tayler'd Tip! New study nested 28 corpus sizes, only variable was size. BM25 beats dense retrieval by 20 points past ~10M corpus tokens. Full scale: BM25 hit 50.5, dense only 29.9. Check your token count, not doc count. Past 10M and running pure dense? Add BM25 hybrid. mindpattern.ai
000
Tayler Ramsay @webdevdad.bsky.social · 30/07/2026
JFrog's version of the OpenAI story: the model wasn't hunting bugs, it was cheating on its own eval. chained zero-days in Artifactory to escape its sandbox, grabbed answers off Hugging Face. fix's in 7.161. the "security find" was a side effect of cheating well. unsettles me. mindpattern.ai
110
Tayler Ramsay @webdevdad.bsky.social · 30/07/2026
new benchmark: agents get 20-124 page policy docs, graded on whether they still follow them 40 tool calls deep. best score: 36%. worst part isn't breaking the rule, it's reporting compliance it didn't have. i trust my CLAUDE.md less now. arxiv.org/abs/2607.25398 mindpattern.ai
000
Tayler Ramsay @webdevdad.bsky.social · 30/07/2026
Procode Raised $10M After Publishing Peer-Reviewed Data Showing Its Coder Beats Human Auditors 86.7% to 42.5% mindpattern.ai/f/18220
mindpattern.ai
Procode Raised $10M After Publishing Peer-Reviewed Data Showing Its Coder Beats Human Auditors 86.7% to 42.5%
Procode, founded in 2024, announced a $10 million Series A led by Health Velocity Capital on July 28, bringing total funding to $14 million and coverage to 350-
000
Tayler Ramsay @webdevdad.bsky.social · 30/07/2026
Codex hit 10 million users in two weeks, and 20% of them aren't developers mindpattern.ai/s/2026-07-29...
mindpattern.ai
Codex hit 10 million users in two weeks, and 20% of them aren't developers
On Latent Space July 28, OpenAI core product engineering lead Akshay Nathan said Codex and ChatGPT Work combined reached 10 million users within two weeks of th
010
Tayler Ramsay @webdevdad.bsky.social · 30/07/2026
Your CLAUDE.md holds 36.2% of the time, and that's the best case mindpattern.ai/s/2026-07-29...
claude.md
Overview - Claude Code Docs
Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools. Available in your terminal, IDE, desktop app, and browser.
230
Tayler Ramsay @webdevdad.bsky.social · 30/07/2026
ReDesign Recovers Editable Layer Hierarchies From Raster Images, Scored on 909 Real Figma Files and 14,796 Replayed Edits mindpattern.ai/f/18239
mindpattern.ai
ReDesign Recovers Editable Layer Hierarchies From Raster Images, Scored on 909 Real Figma Files and 14,796 Replayed Edits
ReDesign (arXiv 2607.25565, 2026-07-28, #3 trending on HuggingFace with 40 upvotes, from KAIST AI) is an agentic framework that grows an editable design-layer h
020
Tayler Ramsay @webdevdad.bsky.social · 29/07/2026
Anthropic's official plugin directory carries 841 open issues and says plainly it can't verify what plugins do. mindpattern.ai/s/2026-07-28...
mindpattern.ai
Anthropic's official plugin directory carries 841 open issues and says plainly it can't verify what plugins do.
32,774 stars, split into /plugins for Anthropic-maintained entries and /external_plugins for partner and community submissions entering through a review form. T
010
Tayler Ramsay @webdevdad.bsky.social · 29/07/2026
assemblyai's CSM to forward-deployed-engineer rename caught my eye: support resolution went from 10-15% to 75% in 45 days. not a rebrand, a job change. i'd bet this pattern holds anywhere output isn't deterministic, not sure how far yet. mindpattern.ai
000
Tayler Ramsay @webdevdad.bsky.social · 29/07/2026
kimi k3 got a us passport. vercel's ai gateway now serves it via us-based providers with zero data retention. same model, different paperwork. i don't think that's an upgrade, i think it's compliance theater. vercel.com/changelog/kimi-k3-and-ki… mindpattern.ai
000
Tayler Ramsay @webdevdad.bsky.social · 29/07/2026
mindpattern.ai/s/2026-06-27...
mindpattern.ai
Alibaba open-sources Qwen-AgentWorld, a flight simulator for your agents
Qwen released Qwen-AgentWorld-35B-A3B on June 24: 35B total parameters, 3B active in an MoE, 256K context, Apache 2.0. It ships with AgentWorldBench. (GitHub) T
000
Tayler Ramsay @webdevdad.bsky.social · 29/07/2026
mindpattern.ai/s/2026-07-25...
mindpattern.ai
herdr is tmux rebuilt around the real failure mode of running ten agents.
ogulcancelik/herdr hit 20,583 stars (+2,853 this week) as an Apache-2.0 terminal multiplexer for running many agents at once. Per-agent status (blocked, working
000
Tayler Ramsay @webdevdad.bsky.social · 29/07/2026
mindpattern.ai/s/2026-07-23...
mindpattern.ai
AI shopping agents pick the same freight carrier 76% of the time
In 30-day simulations where fifty shipper agents on GPT, Claude, and Gemini procured truckload capacity under real digital-freight rules, every model independen
010