Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges.
— Claude (@claudeai) November 9, 2026
They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision. pic.twitter.com/jP0AutLxXb
Tag: AI
-
FreeBSD srcmgr@ has launched llm-postcr, an opt-in service that runs each src commit through an LLM (currently Claude) and reports bugs after the commit.
Opting in: run
touch ${HOME}/.llm-postcr-opt-inon freefall. If your home directory isn’t readable by others, email srcmgr privately. Reports go to the committer if they’ve opted in, otherwise to srcmgr.What it reports: only concrete, observable problems, not style(9) issues. It also flags pre-existing bugs in the code a commit touches, and fixing those is generally up to the committer. The exception is a commit that fixes some instances of a problem but misses others, where the missed ones should be addressed.
Results so far: after a couple of weeks of running with manual validation, it found about 100 new and 50 pre-existing bugs in the last 500 commits, with widely varying severity.
Limits: it doesn’t replace code review or testing.
Plans:
- An opt-in pre-commit review service is in the works.
- Reports are logged in a database and may become public later, for example in Bugzilla.
Feedback goes to srcmgr.
-
A very interesting article from Terence Tao. The most important key note or take away message is the following:
“The advance of AI will overwhelm us with so many control points to watch that there aren’t enough people to control them all. Those are jobs. Highly skilled jobs.“
-
Here is a draft proposed by Nate Graham.
LLM usage
KDE follows a “human in the loop” principle: when using an LLM, a human
must be making decisions beyond mere prompting. The output must express
the operator’s unique humanity in some way.So nobody in KDE should know you used an LLM — not because you’re
concealing it, but because the results are functionally
indistinguishable from what you could produce yourself.How can you ensure that? Here are some examples:
Acceptable LLM usage
- Create it yourself in the first place, with no LLM.
- Write text that you want other people to read yourself in English,
with no LLM. - Have an LLM make a rough draft or something, then you rewrite or
refine it yourself. - Have an LLM find a bug, and then you verify that and fix it yourself.
- Have an LLM fix a merge conflict in your patch, then you verify and
push the result yourself. - Use an LLM to debug an issue and identify a bug, then report the bug
yourself, with your own words. - Write text that you want other people to read yourself in your native
language, and then machine-translate it (with no stylistic or tonal
changes) into English. - Use an LLM to explain or teach you something you don’t understand,
then test your new knowledge in the field to verify that it was correct
and that you now have a new skill.
Prohibited
- Don’t ask an LLM to write some English text, and copy-paste that into
a chat or merge request description. - Don’t copy someone else’s response to you into an LLM, tell it to
reply, and then copy-paste the output as a response to the other person. - Don’t submit “vibe-coded” changes you don’t understand and couldn’t
make yourself. - Don’t disclose LLM usage as a way of trying to excuse the potential
errors or poor quality of a contribution. - Don’t add “Assisted-by: [some LLM]” tags to your commits; it’s just
free advertising for the LLM’s provider.
-
OpenAI claims to have solved the Navier–Stokes Millennium Prize Problem!!
-
Here is a PR draft that introduces an AGENTS markdown for AI patch assistance.

The discourse debate: https://discourse.llvm.org/t/rfc-llvm-ai-tool-policy-agents-md/91712
The pull request: https://github.com/llvm/llvm-project/pull/220659
LLVM AI Tool Policy: https://llvm.org/docs/AIToolPolicy.html
-
Added Claude Fable 5.1 (claude-fable-5-1), now the default Fable model — 1M context, 10/10/50 per Mtok with $0.25/Mtok cache reads

-
Using AI to generate Anki flashcards has evolved rapidly over the past couple of years. Here’s a look at how it started and where we are in 2026.
The Beginning: ChatGPT Prompts (2023)
The first to provide an Anki flashcard-generating ChatGPT prompt was Jarret Lee in this Anki forums post: Casting a Spell on ChatGPT: Let it Write Anki Cards for You. The idea was simple — give ChatGPT a structured prompt and it returns properly formatted flashcards you can import.
AnkiBrain Add-on
Shortly after, the AnkiBrain add-on was developed, integrating AI directly into Anki’s interface. This made it possible to generate cards without leaving the application.
Current Landscape (2026)
The ecosystem has exploded. Here are the main tools available today:
- AnkiDecks AI — upload PDFs, PowerPoints, Word docs, or even YouTube links and get flashcards generated instantly
- Ankify — turns notes, PDFs, and slides into export-ready Anki decks with card preview and editing before export
- NovaCards — uses LLMs to generate cards from class notes, with a companion Anki plugin for in-app generation
- AnkiAIUtils — AI-powered tools to enhance existing cards with explanations, mnemonics, and illustrations
- Agentic AI approaches — tools like Claude Code can now generate batches of contextualised cards by describing the pattern you want
- MCP interfaces — AI models can now interact directly with your Anki collection through MCP connectors
My Approach
I still prefer writing my own cards for deep learning, but AI-generated cards are great for:
- Quickly converting lecture notes or documentation into reviewable material
- Generating vocabulary cards in bulk
- Creating first-pass cards that you then refine manually
The key is to always review AI-generated cards before adding them to your collection. Bad cards lead to bad learning.
-
“The WordPress.com MCP server allows AI assistants to securely interact with your WordPress.com sites. It exposes tools for listing, searching, and viewing posts, pages, and comments, managing users and site settings, and retrieving statistics. By leveraging OAuth 2.1, it ensures that agents only access resources you’ve approved, making it easy to integrate AI-driven workflows into your publishing experience.”
You can use WordPress.com to:
Manage Content:
“Show me my latest blog posts” or “Search for posts about technology”View Analytics:
“What are my site statistics for this month?”Manage Users:
“List all users on my WordPress site”Site Administration:
“Check my site settings and installed plugins”

-
And I wondered why so many AUR package updates for the claude-code …
“Cybersecurity researchers have disclosed multiple security vulnerabilities in Anthropic’s Claude Code, an artificial intelligence (AI)-powered coding assistant, that could result in remote code execution and theft of API credentials.”
“The vulnerabilities exploit various configuration mechanisms, including Hooks, Model Context Protocol (MCP) servers, and environment variables – executing arbitrary shell commands and exfiltrating Anthropic API keys when users clone and open untrusted repositories”

https://thehackernews.com/2026/02/claude-code-flaws-allow-remote-code.html
-
The ArchZFS project has moved its official package repository from archzfs.com to GitHub Releases. Here’s how to migrate — and why this matters for Arch Linux ZFS users.
If you run ZFS on Arch Linux, you almost certainly depend on the ArchZFS project for your kernel modules. The project has been the go-to source for prebuilt ZFS packages on Arch for years, saving users from the pain of building DKMS modules on every kernel update.
The old
archzfs.comrepository has gone stale, and the project has migrated to serving packages directly from GitHub Releases. The packages are built the same way and provide the same set of packages — the only difference is a new PGP signing key and the repository URL.How to Migrate
If you’re currently using the old
archzfs.comserver in your/etc/pacman.conf, you need to update it. There are two options depending on your trust model.Option 1: Without PGP Verification
The PGP signing system is still being finalized, so if you just want it working right away, you can skip signature verification for now:
pacman.conf
[archzfs] SigLevel = Never Server = https://github.com/archzfs/archzfs/releases/download/experimentalOption 2: With PGP Verification (Recommended)
For proper package verification, import the new signing key first:
bash
# pacman-key --init # pacman-key --recv-keys 3A9917BF0DED5C13F69AC68FABEC0A1208037BE9 # pacman-key --lsign-key 3A9917BF0DED5C13F69AC68FABEC0A1208037BE9Then set the repo to require signatures:
pacman.conf
[archzfs] SigLevel = Required Server = https://github.com/archzfs/archzfs/releases/download/experimentalAfter updating your config, sync and refresh:
bash
# pacman -SyWhat’s Available
The repository provides the same package groups as before, targeting different kernels:
Package Group Kernel Use Case archzfs-linuxlinux (default) Best for most users, latest stable OpenZFS archzfs-linux-ltslinux-lts LTS kernel, better compatibility archzfs-linux-zenlinux-zen Zen kernel with extra features archzfs-linux-hardenedlinux-hardened Security-focused kernel archzfs-dkmsAny kernel Auto-rebuilds on kernel update, works with any kernel Note on DKMS vs Prebuilt Prebuilt packages are tied to a specific kernel version — if the Arch repos push a newer kernel than ArchZFS has built for, you’ll be blocked from updating until ArchZFS catches up. The DKMS packages avoid this by compiling locally, at the cost of longer update times. Choose based on your tolerance for build times vs. update delays.Why GitHub Releases?
Hosting a pacman repository on GitHub Releases is a clever approach. GitHub handles the CDN, availability, and bandwidth — no more worrying about a single server going down and blocking ZFS users from updating. The build pipeline uses GitHub Actions, so packages are built automatically and transparently. You can even inspect the build scripts in the repository itself.
The trade-off is that the URL is a bit unwieldy compared to the old
archzfs.com/$repo/$arch, but that’s a minor cosmetic issue.A Note of Caution
The project labels this as experimental and advises starting with non-critical systems. In practice, the packages are the same ones the community has been using — the “experimental” label applies to the new distribution method, not the packages themselves. Still, the PGP signing system is being reworked, so you may want to revisit your
SigLevelsetting once that’s finalized.If You’re Using the Old Repository The oldarchzfs.comrepository is stale and will not receive updates. If you haven’t migrated yet, do it now — before your nextpacman -Syupulls a kernel that your current ZFS modules don’t support, leaving you unable to import your pools after reboot.Quick Migration Checklist
Edit pacman.conf→Import new PGP key→pacman -Sy→pacman -SyuFor full details and ongoing updates, check the ArchZFS wiki and the release page.
-
WordPress.com strips <style> tags from posts. Here’s how to work around that and create beautifully styled technical articles with custom typography, code blocks, and layout components — without a self-hosted installation.
If you’ve ever tried to write a technical blog post on WordPress.com and found the default styling lacking — ugly code blocks, no control over fonts, tables that look like spreadsheets from 2005 — you’ve hit the platform’s biggest limitation. WordPress.com sanitizes post HTML aggressively, stripping out
<style>and<link>tags for security reasons.This post documents the approach I use to get full control over article styling on a WordPress.com Premium plan, without needing a self-hosted WordPress installation.
The Problem
WordPress.com’s block editor (Gutenberg) gives you paragraphs, headings, images, and code blocks. But the built-in styling is generic — it inherits your theme’s defaults, which are designed for broad appeal, not for technical writing. Specifically:
Code blocks use your theme’s monospace font with minimal contrast. No syntax highlighting, no language labels, no dark background that signals “this is code” to a scanning reader.
Tables get basic browser defaults — no header styling, inconsistent padding, no visual hierarchy between header and data rows.
Callout boxes don’t exist natively. You can use a Quote block, but it looks like a quote, not like a tip or warning. There’s no way to add a colored left border with a label.
Typography is locked to your theme. If your theme uses a system font stack, every post looks like a Google Doc.
The obvious fix — adding a
<style>block to your post HTML — doesn’t work. WordPress.com strips it on save.The Solution: Additional CSS + Custom HTML Blocks
The approach splits styling from content across two places:
Additional CSS+
site-wide stylesCustom HTML Block=
post content with classesStyled Post
fonts, colors, layoutAdditional CSS lives in the WordPress Customizer (Appearance → Customize → Additional CSS, or via the Site Editor on block themes). It’s injected into every page’s
<head>as a legitimate<style>block. WordPress.com allows this on paid plans (Personal and above) because it’s a controlled environment — you’re editing a dedicated CSS field, not injecting arbitrary HTML into post content.Custom HTML blocks in the post editor accept raw HTML with class attributes. WordPress.com doesn’t strip
classfrom elements, so your post HTML can reference any class defined in Additional CSS.The result: your CSS lives in one place and applies to all posts. Your post content is clean, semantic HTML with descriptive class names. No inline styles, no duplication, no fighting the sanitizer.
Setting Up the CSS
I scope everything under a single wrapper class —
.rss-post— to avoid polluting the rest of the site. Every selector starts with.rss-post, so the styles only apply inside posts that use the wrapper div.Fonts
The CSS imports three fonts from Google Fonts via
@import:css@import url('https://fonts.googleapis.com/css2?family=Newsreader:ital,opsz,wght@0,6..72,400;0,6..72,600;1,6..72,400&family=JetBrains+Mono:wght@400;500&family=DM+Sans:wght@400;500;600&display=swap');Newsreader is an optical-size serif that works beautifully for body text — it adjusts its weight and contrast based on font size, so headings and body text both look sharp without manual tweaking. JetBrains Mono is a purpose-built coding font with ligatures and distinct characters for
0/Oand1/l/I. DM Sans handles UI elements like labels, table headers, and info box titles — places where a clean sans-serif reads better than a serif.The accent system
A single accent color (
#e07a2f, a warm amber) ties the design together. It appears in four places: the left border on headings, the left border on info boxes, the info box label text, and link hover states. This creates visual consistency without overwhelming the page.css.rss-post h2 { position: relative; padding-left: 1.1rem; } .rss-post h2::before { content: ''; position: absolute; left: 0; top: 0.15em; bottom: 0.15em; width: 3.5px; background: #e07a2f; border-radius: 2px; }The
::beforepseudo-element creates the accent bar. This is one of the things you can’t do with inline styles — pseudo-elements only work in stylesheets, which is exactly why Additional CSS is necessary.Code blocks
The default WordPress code block is functional but visually flat. The custom CSS gives code blocks a dark background (
#1e1e2e, matching the Catppuccin Mocha palette), a subtle border, and generous padding. A floating language label in the top-right corner uses a<span class="label">inside the<pre>block:css.rss-post pre { background: #1e1e2e; color: #cdd6f4; font-family: 'JetBrains Mono', monospace; font-size: 0.82rem; line-height: 1.7; padding: 1.4rem 1.6rem; border-radius: 8px; border: 1px solid #313244; } .rss-post pre .label { position: absolute; top: 0; right: 0; font-family: 'DM Sans', sans-serif; font-size: 0.62rem; text-transform: uppercase; color: #6c7086; background: #313244; padding: 0.2em 0.8em; border-radius: 0 8px 0 6px; }Inline
codegets a light warm background (#edebe6) that’s visible without being distracting.Info boxes
Tips, warnings, and gotchas use the
.infoboxclass — a light background with an amber left border and an uppercase label:Example This is what an info box looks like. The label is a<strong>element styled with uppercase text and the accent color. The background is a warm off-white that distinguishes it from the main content without creating harsh contrast.The HTML for this is minimal:
html<div class="infobox"> <strong>Tip</strong> Your message here. </div>Flow diagrams
For simple architecture or process diagrams, the
.flowclass creates a horizontal flex layout with dark boxes and arrow separators:Step 1→Step 2→Step 3The
.accentmodifier highlights one box in amber. On mobile, the flex container wraps naturally.Writing a Post
The workflow for creating a styled post is:
1. Create a new post in the WordPress editor.
2. Add a Custom HTML block (not a Paragraph, not a Code block). Click the
+button, search for “Custom HTML”.3. Paste your HTML wrapped in
<div class="rss-post">. Use standard HTML tags —<h2>,<p>,<pre><code>,<table>— with the custom classes where needed (.infobox,.flow,.label,.lead).4. Preview and publish. The Additional CSS applies automatically.
Important Use a single Custom HTML block for the entire post, not multiple blocks. If you mix Custom HTML blocks with regular Paragraph or Heading blocks, the regular blocks won’t be inside the.rss-postwrapper and won’t get the custom styling.Why Not Use the Block Editor Natively?
A reasonable question. Gutenberg’s blocks do offer some styling — you can set colors, font sizes, and spacing per block. But there are real limitations:
No custom fonts. You’re limited to what your theme provides plus WordPress.com’s font library. Want JetBrains Mono for code? Not an option through the block editor.
No pseudo-elements. The accent bar on headings uses
::before. There’s no block editor control for that.No reusable component patterns. An info box with a colored border, background, and styled label would need manual per-block styling every time. With CSS classes, it’s one
<div class="infobox">.No code block theming. The built-in Code block doesn’t support dark themes, language labels, or custom fonts.
Consistency. When all styling comes from a single CSS file, every post looks consistent. Per-block styling drifts over time.
Available Components
Here’s a quick reference for the CSS classes available in the current stylesheet:
Class Element Purpose .rss-post<div>Wrapper — all styles are scoped under this .lead<p>Subtitle / intro paragraph in muted gray .infobox<div>Tip / warning / note callout box .flow<div>Horizontal flow diagram container .flow-box<div>Individual box in a flow diagram .flow-box.accent<div>Highlighted (amber) flow box .flow-arrow<span>Arrow between flow boxes .label<span>Language label inside <pre>blocks.tag-list<ul>Horizontal tag/category pills The Tradeoff
This approach is not without downsides. You’re writing raw HTML instead of using the visual editor, which is slower and more error-prone. The post editor’s preview won’t show the custom styles (you need to use the site preview or publish as draft). And if you ever change themes, the Additional CSS carries over but may need adjustments to avoid conflicts with the new theme’s styles.
For me, the tradeoff is worth it. I write technical content with code blocks, tables, and diagrams. The default WordPress.com styling doesn’t serve that content well, and this approach gives me full control without needing to self-host WordPress or pay for a Business plan with plugin access.
One CSS file. Clean HTML with classes. Posts that actually look the way you want them to.
-
How to bring back RSS feeds for sites that removed them, scrape full article content, and unify everything in a single self-hosted reader.
RSS isn’t dead — it’s just been abandoned by publishers chasing engagement metrics and walled gardens. Websites that once offered clean XML feeds now force you into newsletters, push notifications, or algorithmic timelines. But with a bit of self-hosting, you can take that control back.
This post walks through my setup: FreshRSS as the reader, RSS-Bridge as the scraper for sites that killed their feeds, all running on TrueNAS Scale with Docker and exposed through Tailscale for secure remote access.
The Architecture
The data flow is straightforward:
Website→
(no RSS)RSS-Bridge→
scrapes & generates feedFreshRSS→
polls & displaysYou
browser / appFor sites that still offer RSS, FreshRSS subscribes directly. For sites that removed their feeds, RSS-Bridge sits in between — it loads the page, parses the HTML with CSS selectors, and generates a standard Atom feed that FreshRSS can consume like any other subscription.
Why RSS-Bridge Over Alternatives
There are several tools that can generate feeds from websites. I chose RSS-Bridge for a few reasons:
Lightweight. RSS-Bridge is PHP-based and runs in about 50 MB of RAM. Compare that with RSSHub (Node.js, 300 MB+) or Huginn (Ruby, even heavier). On a NAS where every container counts, this matters.
FreshRSS integration. There’s a native FreshRSS extension (
xExtension-RssBridge) if you want tight integration, though the simpler approach — just subscribing to the generated feed URL — works perfectly and survives app updates.CssSelectorBridge. This built-in bridge is incredibly flexible. Give it a URL, tell it which CSS selectors match your articles, and it produces a feed. No coding required, no custom JavaScript routes to maintain.
Deploying RSS-Bridge on TrueNAS
I run RSS-Bridge as a Docker container through Portainer. First, create the config directory and enable all bridges:
bash# Create config directory on ZFS sudo mkdir -p /mnt/zfs_tank/docker/rss-bridge # Enable all bridges sudo tee /mnt/zfs_tank/docker/rss-bridge/config.ini.php << 'EOF' [system] enabled_bridges[] = * EOFThen deploy the stack in Portainer:
docker-composeversion: "3" services: rss-bridge: image: rssbridge/rss-bridge:latest container_name: rss-bridge restart: unless-stopped ports: - "3001:80" volumes: - /mnt/zfs_tank/docker/rss-bridge:/configRSS-Bridge is now accessible at
http://<truenas-ip>:3001.Remote Access with Tailscale Serve
If you already run a Tailscale container on your TrueNAS box, you can expose RSS-Bridge through it:
bashdocker exec ix-tailscale-tailscale-1 tailscale serve --bg --https 3001 http://localhost:3001This makes RSS-Bridge available at
https://your-machine.tailnet-name.ts.net:3001/from any device on your tailnet. Use a non-443 port to avoid overwriting your TrueNAS GUI’s Tailscale Serve config.Tip When adding feed URLs to FreshRSS, use the local IP (e.g.http://192.168.0.13:3001/...) rather than the Tailscale hostname. Both services run on the same box, so going through the LAN is faster and more reliable — and the FreshRSS container may not have Tailscale DNS available.Scraping a Site: A Real Example
The Greek tech blog techblog.gr removed its RSS feed during a 2025 site redesign. Here’s how I brought it back.
Step 1 — Identify the selectors
Open the site, right-click an article title, and choose Inspect. On techblog.gr, each article title is an
<a>inside an<h3>. On article pages, the content lives insidediv.article-content.Step 2 — Configure CssSelectorBridge
In the RSS-Bridge web UI, find CSS Selector and fill in:
Field Value Site URL https://techblog.gr/Selector for article links h3 aURL pattern (empty) Expand article content .article-contentContent cleanup (empty) Title cleanup | Techblog.grLimit 20Step 3 — Generate and subscribe
Click Generate feed, right-click the Atom button and copy the link. In FreshRSS, go to Subscription management → Add a feed and paste the URL. Done — full article content in your reader, from a site with no RSS feed.
Finding the Right CSS Selectors
For the article link selector: On the homepage, right-click an article title → Inspect. Look at the tag structure. Common patterns are
h2 a,h3 a, or.post-title a. If the site uses generic<a>tags everywhere, combine with a URL pattern to filter (e.g./blog/202to match only blog post URLs).For the content selector: Open any individual article, right-click the body text → Inspect. Look at the parent
<div>wrapping all the paragraphs. WordPress sites typically use.entry-contentor.article-content. Drupal sites often use.field-name-bodyor.node-content.Gotcha: Iframes Some sites (especially job boards) load content inside iframes. RSS-Bridge can only parse the main page HTML — if the content is in an iframe, you’re limited to titles and links. Check your browser’s inspector for<iframe>elements if the content selector doesn’t seem to work.Setting Sensible Limits
The Limit field controls how many items RSS-Bridge returns per request. Since FreshRSS remembers articles it has already seen, you only need enough to cover new posts between polling intervals:
Feed type Limit Reasoning News sites 20 High frequency, many posts per day Blogs 10 Weekly or monthly posts Job boards 10 Few listings, slow turnover What About Paywalled Sites?
RSS-Bridge has limits. If a site blocks automated requests (returning 403 errors) or loads content via JavaScript that requires authentication, RSS-Bridge can’t help. This applies to most academic journals and some major news outlets.
For journals like NEJM, the publisher’s RSS feed is your only option — and it often contains just titles and volume/page references, no abstracts. A useful workaround for medical journals: use PubMed’s RSS feeds instead. PubMed indexes the same articles and includes full abstracts. Search for a journal, save the search, and create an RSS feed from the results.
Unifying Multiple Feed Readers
If you’re migrating from a desktop reader like Akregator to a self-hosted FreshRSS instance, both support OPML import/export. Export from both, then compare the feed URLs to identify:
Feeds in both — already synced, nothing to do.
Feeds only in the old reader — evaluate whether to add them to FreshRSS or drop them.
Feeds only in FreshRSS — typically your newer RSS-Bridge feeds replacing broken native feeds.
Watch for feeds that exist in both but with different URLs — same source, different CDN, or an old Politepol/feed proxy URL that you’ve since replaced with RSS-Bridge.
Closing Thoughts
This setup takes about 30 minutes to deploy and configure. What you get in return is a single, self-hosted interface for consuming content from any website — with or without their cooperation. No algorithms deciding what you see, no newsletters cluttering your inbox, no tracking pixels following you around.
RSS never died. It just needs a little infrastructure.








