← Amplification
Amplification · No. 13

The one sentence that makes Fable 5.2 refactor, a Brotli bug under active attack, and the Gloverall duffle

Today: a one-sentence prompt change Anthropic measured at roughly 5x on refactoring, a Brotli flaw whose federal deadline passed yesterday, and the duffle coat.

AI — Build With It

Every issue →

Tell Fable 5.2 the codebase is its own, and it starts merging duplicated code

Anthropic’s prompting docs for Claude Fable 5.2 say that without ownership language, the model extends existing structure even when that duplicates code. Add the sentence “This codebase is yours: what it’s for and how it’s built are your call, and you’ll keep building on it,” and on small coding tasks it merged repeated code in about half of tries, against under a tenth without it. Those are Anthropic’s numbers on Anthropic’s tasks. Fable 5.2 is also priced at $8/$40 per million tokens, a 20% cut from 5.1’s $10/$50 (llm-anthropic 0.31: https://simonwillison.net/2026/Nov/9/llm-anthropic/).

WHAT I’D DO Paste the sentence into the top of your CLAUDE.md or system prompt today (two minutes), then check the next three diffs for consolidated helpers versus new copies. If you want a real answer, see Build This below.

Chrome’s Modern Web Guidance gives your coding agent current web-platform knowledge, and can scan for legacy patterns

It is an Apache-2.0 set of skills from the Chrome, Edge, Firefox and MDN teams: 250 features and 203 use cases covering things like View Transitions, container queries, :has(), the Popover API and Temporal, written as token-light markdown guides, with plugin folders for Claude, Codex, Cursor and Gemini. You can try it without installing: npx modern-web-guidance@latest search “animate a dialog modal backdrop”, or npx modern-web-guidance@latest legacy-scan . to flag legacy patterns in a repo.

WHAT I’D DO Run legacy-scan on one frontend repo (ten minutes) and read the findings before installing anything. If the hits are real, run the install wizard (npx modern-web-guidance@latest install) so agents stop reaching for 2019 patterns.

Gemini 4 Carbon is generally available at $2/$10, and the price doubles on December 1

Per Simon Willison, Gemini 4 Carbon moved from a cyber-defender preview to general availability at $2 per million input tokens and $10 per million output, scheduled to double to $4/$20 on December 1, matching Opus 5.5. It has low, medium and high effort settings, with medium the default. Separately, Muse Ember 1.0 is $0.10/$0.50 for short prompts, the same as GPT-6 Luna and Claude Haiku 5.5, rising to $0.30/$1.50 above 200k tokens (https://simonwillison.net/2026/Nov/10/muse-ember/).

WHAT I’D DO If you want Carbon on your own workload, run the comparison before December 1 while it is half the price. An hour with llm-gemini 0.36 and ten real prompts will tell you whether the later price is worth it.

Node2Shell: a Brotli decoder flaw is under active attack, and the federal patch deadline was yesterday

CVE-2026-111497 is an out-of-bounds write in the Brotli decoder, reachable by sending a crafted Brotli-compressed stream. Unauthenticated attackers can crash a service, leak memory, or in some cases get remote code execution. Microsoft shipped an emergency .NET update on Oct 26, CISA added it to the KEV catalog Oct 27, and agencies had until Nov 10 under BOD 26-04. BleepingComputer gives no CVSS score and no version numbers, so check your vendor advisory for those.

WHAT I’D DO Patch the runtime, restart the processes, and rebuild container images, since a base image carries its own Brotli. If you cannot patch yet, block Brotli-compressed requests at the proxy or disable Brotli decoding.

Protecting SSH keys with the Mac Secure Enclave, without a YubiKey

On macOS Sequoia or later, sc_auth create-ctk-identity -l “ssh-MacBook” -k p-256-ne -t bio creates a non-exportable P-256 key that needs Touch ID for each signature; ssh-keygen -w /usr/lib/ssh-keychain.dylib -K -N “” writes the handle files and SSH_SK_PROVIDER points ssh at it. The author hit enrollment error -25293 until re-enrolling Touch ID, and the prompt reads “ctccardtoken needs to authenticate.” The key cannot be backed up, so register a second key on each server. It sat at 229 points on Hacker News.

Worth Knowing

Every issue →

A 66-pound rodent from 34 million years ago, identified from two teeth

Yakamys sirindhornae, from a lignite mine in Krabi, Thailand, is estimated at about 30 kg (roughly 66 lb), around twice the mass of other rodents of its period. It suggests giant rodents appeared earlier, and in Southeast Asia, than the South American story implies. The mass estimate comes from tooth measurements across seven methods, and the authors call it a ballpark. Published in Scientific Reports.

DINNER-TABLE VERSION The capybara is not the first giant rodent, just the one that survived. This one is known from a lower molar and an incisor.

NASA’s four-person Mars simulation crew is out after more than a year inside

The CHAPEA crew finished on Oct 31 after more than a year in a simulated habitat, growing crops, rationing supplies, running simulated spacewalks and working through communication delays with Earth. NASA says the lessons will inform its planned Moon Base and future Mars missions. The release gives no quantitative results yet.

DINNER-TABLE VERSION The hard part being studied is mostly four people and a delayed phone line, not the hardware.

The Inner Work

Every issue →

Feeling safe predicts how a team learns better than how it performs

Sanner & Bunderson (2015), Organizational Psychology Review, doi:10.1177/2041386614565145. Meta-analysis of 51 studies, about 3,700 teams; the task-complexity moderators were coded from occupational databases rather than measured in the studies, so read that part as suggestive.

Psychological safety is the belief that you can raise a doubt or a mistake without being punished for it. Across 51 studies and about 3,700 teams, Sanner and Bunderson found it correlates with team learning at .58 and with performance at only .32 (learning to performance is .39). Task complexity alone explained roughly a third of the variance in the safety-learning link and nearly half of the safety-performance link: safety pays most where the work is complex enough that someone needs to say “I think this is wrong.” On routine work it buys you comfort and not much else. For an engineering team that means safety is the precondition for review and postmortems working, not a morale perk.

TRY THIS TODAY In your next review or standup, go first: name one mistake or open doubt of your own from this week, in one sentence, before asking anyone else for theirs. Then notice whether the next person’s answer got more specific.

The ownership-sentence A/B

Effort: 30 minutes · Run: once / on demand

Anthropic reports that one sentence moves duplicate-code merging from under 10% of tries to about half, on its own small tasks. Whether that holds on your repo, with your conventions, is a 20-run question a script can answer faster than a week of eyeballing diffs. The point is to get a rate for your code and not an anecdote.

THE PATTERN A vendor-reported prompt effect is a claim about the vendor’s tasks. A paired, repeated run on your own code is cheap, and it applies to any prompt tip you are about to adopt.
How to build it
  1. Pick a small repo, or a branch of one, with two near-identical helpers in it, and write one refactor task prompt that mentions the feature both serve. Commit that as the baseline.
  2. Write a Python script that loops over two conditions (the prompt alone, the prompt plus the ownership sentence) times 10 runs. Each run copies the repo to a temp directory and calls claude -p with the prompt, then saves git diff.
  3. Score each run mechanically: did the two helpers become one (count the function definitions with grep), and do your tests still pass?
  4. Print a table of consolidation rate, test-pass rate and diff size per condition. If the gap is under a few runs, you have your answer: the sentence is not doing much for you.

Dress Better

Every issue →

The duffle coat

A wool duffle is a thick, felted-looking coat with a hood, wooden toggles on rope loops, and a patch or two of leather, cut about mid-thigh. Gloverall’s Original Monty is the reference model, named for Field Marshal Montgomery. It reads as warm and slightly boyish over a rollneck or a shirt and knit, and it is the coat for a cold Wednesday when a navy overcoat feels too office. Wear it with dark jeans or flannel trousers and boots, and leave the hood down in town.

WHAT’S SHIFTING Fall menswear overviews, including The Impression’s, describe a return to clean lines and structure. A duffle has a stiff, defined shape, which fits that better than a slouchy parka does.
WHERE TO BUY Gloverall Original Monty, £437.75 at 15% off the £515 regular price, in eight colours, from the brand: https://www.gloverall.com/products/original-monty-duffle-coat. Gloverall for J. Press at Buck Mason (price not confirmed): https://www.buckmason.com/for-ai/mens/outerwear/coats/duffle-coats/j-press-x-buck-mason-gloverall-monty-duffle-coat. Son of a Stag carries the navy Original (price not confirmed on the brand’s own page): https://sonofastag.com/products/gloverall-original-monty-duffle-coat-navy
Shoulder The seam should land on the bone at the top of your arm; a duffle is bulky already, and a dropped shoulder makes it look borrowed.
Length The hem should sit at mid-thigh, below the back pocket of your jeans, and not at the knee.
Sleeve It should cover the wrist bone with a sweater underneath, so size to the sweater you will actually wear.
Toggles Do the top two toggles up and leave the lower ones open to sit down on the subway without the coat pulling across your thighs.
Where to buy
THE PICK

Trinidad Golden Place

Crown Heights · Trinidadian bakery · Bring cash

A Trinidadian bakery at 788 Nostrand Ave that The Infatuation scored 8.8 (the review is from January 2022). The reason to cross the borough is the goat roti, about $14, and the doubles, about $2, which are spicy chickpeas in two layers of soft dough. The line moves fast, whatever the hour.

HOW TO PLAY IT Walk in, expect a line on a Wednesday or Thursday too, and know your order before the counter. Get the goat roti and a doubles for eating there, and a currant bun to carry home. Bring cash. Prices are from The Infatuation’s cheap-eats guide.
ALSO ON THE RADAR

Nepali Bhanchha Ghar

Jackson Heights · Nepali · BYOB

Order the beef jhol momo (about $12), momos in a sesame-and-tomato broth, with sel roti ($2). Bring your own drink.

Lechonera La Isla

East Harlem · Puerto Rican · Counter service

A tiny counter spot at 254 E 125th St; the pernil plate with arroz con gandules is about $16. A 2021 review says it had been around for more than 30 years.