Agent Skills becomes an open standard next to MCP, while developers laugh at AI coding culture

✕Fri 23 Oct ‹›

AI security · Fri 23 Oct

AI found a critical hole in PHP, and the coordinated patch is out: update now

PHP 8.xevery supported version needed the patch

Two memory-corruption bugs in the PHP engine, nicknamed PHP2Shell, were found separately by a researcher using OpenAI's the Planet program and by Wiz Research using . Researchers showed remote code execution on WordPress, Drupal and Joomla before login. After about two months of coordination, patched PHP 8.x releases shipped on Oct 21, a day early because the CVE IDs leaked.

Key points

The patches cover the whole PHP 8.x line; Cloudflare shipped stopgap firewall rules for every plan while sites update.
The release was pulled forward after the CVE IDs leaked with a wrong severity score (10.0 instead of 9.2), and a mistakenly published proof-of-concept video sped up exploit work.
OpenAI published a Codex that walks site owners through checking exposure, a helper rather than a substitute for vendor advisories.
Credit is disputed, and viral outage claims are uncorroborated: "All high profile cyber disclosures are at least a little bit of marketing."

Why it matters: Any client site on WordPress, Drupal or Joomla should be patched now, and asking a coding agent to scan your own projects for affected dependencies is a quick first check.

Sources (3)

[AINews] OpenAI's Patch the Planet program is 1 of 2 parties to discover PHP2Shell vulnerabilities in core PHP · Latent Space
Today's PHP Security Release, and How We're Strengthening PHP Security · Simon Willison
Claude's personal agent · Ben's Bites

Next · Personal agentsPersonal agents are taking off, and the first leaks show why sharing settings matter →

✕Fri 23 Oct ‹›

Personal agents · Fri 23 Oct

Personal agents are taking off, and the first leaks show why sharing settings matter

10MUS Muse downloads in 40 days

Meta's Muse passed 10 million US downloads in 40 days, but researchers found thousands of shared conversations through guessable share links and Meta pulled the feature. Anthropic's Claude Duet is in US-only preview and Gemini in Android Auto now taps your Google data. Riley Brown argues the best use of agents is removing clutter rather than creating it.

Key points

About 6,800 Muse shared-conversation links were enumerable because the URLs were guessable; Meta removed sharing while it hardens it.
Gemini in Android Auto can now use your inbox and other Google data, for example to check whether you'll make a flight.
Examples shown: debloating smart TVs and an e-reader with Claude Code, letting an agent archive obvious spam, and replacing single-purpose apps with a prompt plus a spreadsheet.
A Muse automation that logs new followers to a spreadsheet and flags relevant ones is a simple pattern for social monitoring.

Why it matters: Before rolling an assistant out to a team, check how its sharing links and data access work; the useful wins are small automations that remove busywork.

Sources (3)

How AI is Changing What Work Means · The AI Daily Brief
How Smart People Are Using AI Right Now (7 Insane Use Cases) · Riley Brown
Claude's personal agent · Ben's Bites

Next · AI for scienceClaude-designed proteins bound scorpion toxins in the lab, but the slow part is still the lab →

✕Fri 23 Oct ‹›

AI for science · Fri 23 Oct

Claude-designed proteins bound scorpion toxins in the lab, but the slow part is still the lab

17 of 211AI-designed binders confirmed in the lab

A research firm used Claude with 300 to generate 318,000 protein-binder candidates, tested 211, and an independent lab confirmed 17 bind to four scorpion neurotoxins. None has been shown to neutralise venom yet, and critics note binders were shown before. OpenAI is also pushing into life sciences by acquiring Orchestra Bio for its GPT Rosalind tools.

Key points

17 of 211 Claude-designed binders met the experimental binding criteria.
Dry-lab design took about 64 hours, while lab results took 39 days and a release would take 4 to 7 years.
OpenAI's GPT Rosalind and Workbench agents are positioned as tools for existing pharma companies.
AI lets firms do research outside their field that can sharpen their core work.

Why it matters: AI speeds up the thinking step, but timelines are still set by the physical steps; plan projects around the real bottleneck.

Sources (2)

How AI is Changing What Work Means · The AI Daily Brief
Claude's personal agent · Ben's Bites

Next · Models & infrastructureForecasting models, inference basics and elastic GPUs: what actually runs AI →

✕Fri 23 Oct ‹›

Models & infrastructure · Fri 23 Oct

Forecasting models, inference basics and elastic GPUs: what actually runs AI

~25%lower error than a seasonal baseline in the transit test

Google's TimesFM 3.0 forecasts well out of the box but can only extrapolate levers that have moved in your history. IBM explains why serving a model costs more than training it over time. Modal's CEO argues demand is spiky, so capacity should be pooled and elastic, with for coding agents now a big part of its business.

Key points

TimesFM 3.0 handles holidays and promotions inside the model, but mis-judged a 'double the marketing spend' what-if; 3.0 weights are non-commercial to self-host while 2.5 is Apache 2.0.
Prefill reads the whole prompt and is compute-bound; decode writes one at a time and is memory-bound. They map to time-to-first-token and inter-token latency.
Inference engines like vLLM, Ollama and llama.cpp mostly expose OpenAI-compatible endpoints, so models can be swapped without rewriting code.
Modal's fast startup comes from GPU memory snapshots, checkpointing and a custom filesystem.

Why it matters: When a client asks for AI forecasts or self-hosted models, check the license and whether their history contains the decision they want to test.

Sources (3)

Google's TimesFM 3.0: Is This The End of Traditional Forecasting? · Prompt Engineering
What Is LLM Inference? How AI Models Generate Responses · IBM Technology
Why AI Compute Comes in Rectangles — Erik Bernhardsson, Modal CEO · Latent Space

Next · Standards & dev cultureAgent Skills becomes an open standard next to MCP, while developers laugh at AI coding culture →

✕Fri 23 Oct ‹›

Standards & dev culture · Fri 23 Oct

Agent Skills becomes an open standard next to MCP, while developers laugh at AI coding culture

Anthropic donated Agent to the Agentic AI Foundation, the Linux Foundation group that already hosts MCP. Fireship's satire of 2026 coder types lands on real problems: insecure vibe-coded apps and maintainers swamped by AI pull requests. A Latent Space interview argues small AI teams have to own everything themselves.

Key points

The Agentic AI Foundation was co-founded with Block and OpenAI, so skills and MCP now sit under one neutral home.
Simon Willison found his own earlier prediction of this by asking Claude to search his blog through datasette-mcp.
Security basics still matter for AI-built apps: hard-coded on the front end are a recurring failure.
"You are the team": trusted sources, solid document extraction and meeting users before hiring matter most.

Why it matters: Skills written today should keep working across tools as the standard spreads, which makes them a safer investment for client playbooks.

Sources (4)

Donating Agent Skills to the Agentic AI Foundation · Simon Willison
Quoting Armin Ronacher · Simon Willison
The last real programmers are dying out... · Fireship
You are the Team — Anekha Sokhal, Moshi · Latent Space

Next · AI security & trustYour AI agents are now the easiest way in. Security teams say: verify, then trust →

✕Thu 22 Oct ‹›

AI security & trust · Thu 22 Oct

Your AI agents are now the easiest way in. Security teams say: verify, then trust

84%of ransomware attacks had AI help

Two security briefings this week reach the same conclusion. Attackers now use AI in most ransomware attacks, and they increasingly go through the and identity systems companies have set up themselves. Meanwhile new AI decision tools return confident answers that are hard to audit, and a confident wrong answer is harder to catch than an obvious error.

Key points

A survey of 1,200 IT professionals found 84% of ransomware attacks had AI help. 44% abused a company’s own agents or identity systems.
Old risks still hurt: hacked supermarket fridges ran old firmware even though a fix already existed.
Confidence is not evidence. A model saying “97% sure” can still have been fed altered logs or poisoned data.
The fix is boring and works: give agents their own identities, log everything, use a kill switch, split big decisions into small checkable steps, and keep a human on high-risk actions.

Why it matters: Every Notion agent and AI workflow we build for clients needs access rules, audit logs and an off switch from day one. That is a clear, sellable part of an engagement.

Sources (2)

The patching gap: Hacked fridges, AI ransomware and Mythos finds · IBM Technology
The Problem with Jev: Can You Trust AI Decisions? · IBM Technology

Next · AI agents at workPersonal agents are getting polished and plugged into your apps, but they still work alone →

✕Thu 22 Oct ‹›

AI agents at work · Thu 22 Oct

Personal agents are getting polished and plugged into your apps, but they still work alone

3 tierschatbot → DIY agent → personal agent

A hands-on review of Claude Duet, an explainer on and a demo of Claude’s live dashboards all point the same way. Agents now connect to your email, calendar and finance tools and keep a memory in plain text files. The gaps are speed, being proactive and working as a team.

Key points

There are three tiers: chatbots (no memory), DIY agents that you shape with plain-text rule and memory files, and ready-made personal agents that run all day in the cloud.
Claude Duet inherits your plugins and connectors and looks polished, but parallel tasks took 10+ minutes and it can’t be shared with a team.
Live dashboards pull fresh data through connectors and can be shared by link, so a short prompt can replace a static CSV report.
The trade-off is lock-in: the more an agent knows about you, the harder it is to switch.

Why it matters: Clients will ask “personal agent or DIY setup?” Notion-based memory and rules files are a portable middle path that avoids lock-in.

Sources (3)

I tested Claude Duet. Is it worth switching? · How I AI
Personal Agents, Clearly Explained · Jeff Su
Claude’s New Live Dashboards Are Amazing · Skill Leap AI

Next · Open models & costWhy AI keeps getting cheaper: memory is the bottleneck, and the fixes are going open source →

✕Thu 22 Oct ‹›

Open models & cost · Thu 22 Oct

Why AI keeps getting cheaper: memory is the bottleneck, and the fixes are going open source

155Ttokens a day served by DeepSeek’s now-open engine

DeepSeek open-sourced the engine it uses to serve 155 trillion a day. An IBM explainer shows why that design works: writing an answer is limited by memory speed, not raw compute. And a local-AI build shows a tuned open model on a laptop can handle most everyday tasks.

Key points

Reading a prompt is compute-heavy, but writing the answer waits on memory. DeepSeek runs the two on separate machines.
DeepSeek keeps prompt caches on SSD for days, and 96% of input tokens hit the cache. Keep your prompt prefix stable to benefit.
Since 2006, chip compute has grown about 6,500x while memory bandwidth grew far more slowly. Batching many requests together is the main workaround.
Local models are “good enough” for about 95% of tasks (a creator’s claim). A 180B open model ran at about 72 tokens/s on a laptop.

Why it matters: For privacy-sensitive clients, a hybrid setup (local model for routine work, cloud for hard tasks) is now realistic. Stable prompt templates also cut API costs.

Sources (3)

DeepSeek Just Open-Sourced How It Serves 155T Tokens a Day · Prompt Engineering
The AI Memory Wall: Why LLMs & AI Accelerators Hit a Bottleneck · IBM Technology
Stop paying for AI, run everything locally (here’s how) · David Ondrej

Next · The model raceBenchmarks aren’t the whole story: AI “personality” now decides which model people use →

✕Thu 22 Oct ‹›

The model race · Thu 22 Oct

Benchmarks aren’t the whole story: AI “personality” now decides which model people use

900questions in the new AI personality benchmark

Google’s new Gemini 4 Argon tops several leaderboards but still trails rivals at coding. A new personality finds it scores highest for “slop” (filler and generic phrasing). As top models get equally smart, tone and style are becoming the real difference.

Key points

Gemini 4 Argon is #1 on several leaderboards but loses coding benchmarks to Opus 5.5 and GPT-6 Astra.
The AA Personality Bench uses 900 questions across 30 traits. Muse leads on warmth, Opus 5.5 on imagination, and Argon on slop.
In a hands-on test, code-generated motion graphics beat native AI video and were easier to edit.
Policy note: a US federal memo bans DeepSeek models from civilian government systems.

Why it matters: Choose models per task, not by leaderboard. For client-facing assistants, test tone and filler, not just accuracy.

Sources (2)

Does Personality Matter More Than Intelligence in AI? · The AI Daily Brief
Did Google just dethrone OpenAI and Anthropic? · Fireship

Next · AI in AsiaFrom Changi’s runways to China’s robot schools, Asia is putting AI to work in the physical world →

✕Thu 22 Oct ‹›

AI in Asia · Thu 22 Oct

From Changi’s runways to China’s robot schools, Asia is putting AI to work in the physical world

−90%false alerts on Changi’s runways after the AI upgrade

Two reports show AI leaving the screen. Singapore’s Changi Airport uses AI cameras to spot runway debris as small as 4 cm. In China, “robot schools” train on everyday tasks, at a scale no other country matches.

Key points

At Changi, cameras on 12 towers watch each 4 km runway. The AI upgrade cut false alerts by 90% and improved detection sevenfold.
A person still checks every alert before a team goes out, usually within 1 to 5 minutes.
China made about 90% of humanoid robots shipped in 2025 (analyst figures). Training centres teach robots to cook and run machinery.
Risks: a slowing economy and tighter controls may hold back research.

Why it matters: These are strong, local examples for workshops: AI flags, a human verifies, a team acts. It’s the same pattern we design in Notion workflows.

Sources (2)

How Changi Airport Finds Tiny Objects On Its Runways With AI · CNA Insider
Inside China’s Robot Schools: How Embodied AI Is Being Built · CNA Insider

Friday 23 Oct

Thursday 22 Oct

AI found a critical hole in PHP, and the coordinated patch is out: update now

Key points

Personal agents are taking off, and the first leaks show why sharing settings matter

Key points

Claude-designed proteins bound scorpion toxins in the lab, but the slow part is still the lab

Key points

Forecasting models, inference basics and elastic GPUs: what actually runs AI

Key points

Agent Skills becomes an open standard next to MCP, while developers laugh at AI coding culture

Key points

Your AI agents are now the easiest way in. Security teams say: verify, then trust

Key points

Personal agents are getting polished and plugged into your apps, but they still work alone

Key points

Why AI keeps getting cheaper: memory is the bottleneck, and the fixes are going open source

Key points

Benchmarks aren’t the whole story: AI “personality” now decides which model people use

Key points

From Changi’s runways to China’s robot schools, Asia is putting AI to work in the physical world

Key points

Send us feedback