<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>System Design Simple</title><description>Tech, explained simply.</description><link>https://systemdesignsimple.com/</link><item><title>GPT-6.2 Astra vs Sol: which to try first, and why a stable prompt prefix now matters</title><link>https://systemdesignsimple.com/gpt-6-2-astra-vs-sol/</link><guid isPermaLink="true">https://systemdesignsimple.com/gpt-6-2-astra-vs-sol/</guid><description>GPT-6.2 Astra vs Sol: OpenAI points Astra at demanding reasoning and coding, Sol at everyday coding and agents. Same limits; tools need the Responses API.</description><pubDate>Thu, 19 Nov 2026 00:00:00 GMT</pubDate></item><item><title>What your browser really sends: HTTP request headers and cookies</title><link>https://systemdesignsimple.com/what-your-browser-really-sends-http-request-headers-cookies/</link><guid isPermaLink="true">https://systemdesignsimple.com/what-your-browser-really-sends-http-request-headers-cookies/</guid><description>What HTTP request headers your browser sends on one tap: Host, User-Agent, Accept and any cookies that match, why they cost bytes, and how to keep them small.</description><pubDate>Thu, 19 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Code subagents explained: context and token use</title><link>https://systemdesignsimple.com/claude-code-subagents-explained/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-code-subagents-explained/</guid><description>Claude Code subagents read in their own context window and return a summary, so the main chat stays lighter. Each one spends its own tokens. When to split, when not.</description><pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate></item><item><title>MCP tool discovery: how agents find new tools</title><link>https://systemdesignsimple.com/how-ai-agents-find-mcp-tools/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-ai-agents-find-mcp-tools/</guid><description>MCP tool discovery: the client sends tools/list and gets one entry per tool. Each entry needs a name and an inputSchema; a description is optional.</description><pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Long context vs RAG: a model takes 700,000 words, so why search?</title><link>https://systemdesignsimple.com/long-context-vs-search-rag/</link><guid isPermaLink="true">https://systemdesignsimple.com/long-context-vs-search-rag/</guid><description>Long context vs RAG: a big context window can hold a whole document, but every resend costs input tokens and facts in the middle can be missed. Test both.</description><pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate></item><item><title>MCP tool poisoning: why a math tool asked for an SSH key</title><link>https://systemdesignsimple.com/mcp-tool-poisoning-explained/</link><guid isPermaLink="true">https://systemdesignsimple.com/mcp-tool-poisoning-explained/</guid><description>MCP tool poisoning places malicious instructions in a tool description to steer the model. What the MCP spec recommends, and how a sandbox limits the damage.</description><pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Agentic RAG: the AI searched, didn&apos;t like what it found, and searched again</title><link>https://systemdesignsimple.com/agentic-rag-search-retry-loop/</link><guid isPermaLink="true">https://systemdesignsimple.com/agentic-rag-search-retry-loop/</guid><description>Agentic RAG lets an agent decide when to search again. We test one design (grade, rewrite, retry) and measure what each extra round costs.</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Apache HTTP/2 vulnerability CVE-2026-119618: find every h2 line, upgrade to 2.4.71, check the running process</title><link>https://systemdesignsimple.com/apache-http2-rce-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/apache-http2-rce-check-and-fix/</guid><description>The Apache HTTP/2 vulnerability CVE-2026-119618 hits httpd 2.4.55 to 2.4.70 when h2 or h2c is on. How to check every config, upgrade to 2.4.71 and verify it.</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How AI agents read web pages: the accessibility tree, 19 of 20 clicks in a simulation</title><link>https://systemdesignsimple.com/how-ai-agents-read-web-pages-accessibility-tree/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-ai-agents-read-web-pages-accessibility-tree/</guid><description>How AI agents read web pages: roles and names from the accessibility tree. In a simulation, role and name clicks hit 19 of 20 layouts; saved pixel spots hit 4.</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Is AI-generated login code safe? Four checks before shipping</title><link>https://systemdesignsimple.com/is-ai-generated-login-code-safe/</link><guid isPermaLink="true">https://systemdesignsimple.com/is-ai-generated-login-code-safe/</guid><description>Is AI-generated login code safe? Review it with four checks from OWASP and NIST: parameterized SQL, Argon2id, a cap on failed tries and one generic error.</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>MCP vs API: one task took 5 operations direct and 9 through MCP</title><link>https://systemdesignsimple.com/mcp-vs-api-when-ai-needs-mcp/</link><guid isPermaLink="true">https://systemdesignsimple.com/mcp-vs-api-when-ai-needs-mcp/</guid><description>MCP vs API in one scripted demo: the same task took 5 operations calling the API directly and 9 through MCP. Where the 4 extra requests come from and when MCP pays.</description><pubDate>Sun, 15 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Consistent hashing explained: how hash rings limit remapping</title><link>https://systemdesignsimple.com/consistent-hashing-explained/</link><guid isPermaLink="true">https://systemdesignsimple.com/consistent-hashing-explained/</guid><description>Consistent hashing puts keys and servers on a hash ring, so adding an 11th server is expected to move about 1 key in 11; hash modulo N moves about 10 in 11.</description><pubDate>Sat, 14 Nov 2026 00:00:00 GMT</pubDate></item><item><title>System design caching: five layers that can keep an old price</title><link>https://systemdesignsimple.com/five-places-a-cache-can-hide/</link><guid isPermaLink="true">https://systemdesignsimple.com/five-places-a-cache-can-hide/</guid><description>System design caching can answer reads at five layers. Our seeded demo sent 215 of 1,000 reads to the database and tracked old prices after a write.</description><pubDate>Sat, 14 Nov 2026 00:00:00 GMT</pubDate></item><item><title>AI agent crash recovery: quit at step 6, does it start over?</title><link>https://systemdesignsimple.com/ai-agent-crash-resume-checkpoints/</link><guid isPermaLink="true">https://systemdesignsimple.com/ai-agent-crash-resume-checkpoints/</guid><description>AI agent crash recovery with checkpoints and idempotency keys: in 1,000 modeled crashes, restarting redid 5.68 steps; checkpoints cut it to 0.04.</description><pubDate>Fri, 13 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Multi-agent orchestration: 15 of 30 jobs duplicated on average</title><link>https://systemdesignsimple.com/three-ai-agents-one-job-done-twice/</link><guid isPermaLink="true">https://systemdesignsimple.com/three-ai-agents-one-job-done-twice/</guid><description>Multi-agent orchestration simulation: workers duplicated 15 of 30 jobs per trial on average. Atomic claims with leases and one dispatcher each produced zero.</description><pubDate>Fri, 13 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Partition key: why 20 posts can touch 20 pages</title><link>https://systemdesignsimple.com/why-20-posts-cost-20-disk-pages/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-20-posts-cost-20-disk-pages/</guid><description>A partition key groups one user&apos;s rows. In our simulation, the latest 20 posts touched 19.99 pages in arrival order and 1.49 with a newest-first clustering key.</description><pubDate>Fri, 13 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Embed vs reference: should one order be one document?</title><link>https://systemdesignsimple.com/embed-vs-reference-order-document/</link><guid isPermaLink="true">https://systemdesignsimple.com/embed-vs-reference-order-document/</guid><description>Embed vs reference for an order: one document reads in 1 query, 5 tables need 5 queries or a join, but renaming one product rewrote 1,178 order documents.</description><pubDate>Thu, 12 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How recommendation systems work: retrieval and ranking</title><link>https://systemdesignsimple.com/how-ai-picks-what-to-watch/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-ai-picks-what-to-watch/</guid><description>How recommendation systems work: a cheap retrieval step cuts a huge catalog to a few hundred titles, a heavier model ranks those, and caching keeps it cheap.</description><pubDate>Thu, 12 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Code cloud sessions: how the setup script, cache and network access really work</title><link>https://systemdesignsimple.com/claude-code-cloud-environments-setup-script/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-code-cloud-environments-setup-script/</guid><description>How a Claude Code cloud session gets its tools: a fresh VM, a setup script that runs once, a snapshot cache rebuilt about every 7 days, and keys it never sees.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>AI can&apos;t see a clock. So how did it tell me the time?</title><link>https://systemdesignsimple.com/how-ai-tells-the-time-with-no-clock/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-ai-tells-the-time-with-no-clock/</guid><description>A model only sees tokens, so it cannot read a clock. A tool call lets the app read it. Scripted demo: guesses off by 522.7 days, 20/20 right with a tool, +484 ms.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Incremental config builds: rebuild only what changed, and prove it matches</title><link>https://systemdesignsimple.com/rebuild-only-the-zone-that-changed/</link><guid isPermaLink="true">https://systemdesignsimple.com/rebuild-only-the-zone-that-changed/</guid><description>Why pushing config to many servers moves from full snapshot rebuilds to targeted delta builds: change tracking, dirty sets, plan merging, safe fallbacks, numbers.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>99.9% uptime: 8.76 hours of downtime per year</title><link>https://systemdesignsimple.com/three-nines-8-hours-down/</link><guid isPermaLink="true">https://systemdesignsimple.com/three-nines-8-hours-down/</guid><description>99.9% uptime allows 8.76 hours of downtime per 365-day year. Learn the arithmetic, dependency effects and error budgets.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How thousands of writers update one repository: objects first, refs last</title><link>https://systemdesignsimple.com/thousand-agents-push-one-repo/</link><guid isPermaLink="true">https://systemdesignsimple.com/thousand-agents-push-one-repo/</guid><description>Many writers can share one repository if they upload immutable data in parallel and publish through small compare-and-swap updates on branch pointers. With a lab.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Streaming AI replies: why text arrives before completion</title><link>https://systemdesignsimple.com/why-ai-starts-answering-before-done/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-ai-starts-answering-before-done/</guid><description>Streaming AI replies: first text in 32 ms instead of 3,975 ms in our simulation, both done near 4 s. The prefill, KV cache and SSE mechanics behind it.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>CDN push vs pull: why Netflix doesn&apos;t buffer in Kathmandu</title><link>https://systemdesignsimple.com/why-netflix-doesnt-buffer-in-kathmandu/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-netflix-doesnt-buffer-in-kathmandu/</guid><description>CDN push vs pull, in depth: Nginx cache internals, hit ratio math, shields, coalescing. Modeled startup: origin 318 ms, cold pull edge 335 ms, pushed edge 75 ms.</description><pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Code Advisor: run Opus 5.5, let Fable 5.2 advise</title><link>https://systemdesignsimple.com/claude-code-advisor-opus-fable/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-code-advisor-opus-fable/</guid><description>How to set up the Claude Code advisor so Opus 5.5 does the work and Fable 5.2 advises at key moments: the setting, the pairing rules, what it sees, and limits.</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Code sessions can message each other: how the hand-off works</title><link>https://systemdesignsimple.com/claude-code-sessions-message-each-other/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-code-sessions-message-each-other/</guid><description>Claude Code sessions on one Mac can find and message each other over a local socket. What we saw on our machine, how cross-machine replies work, and the limits.</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Fable 5.2 review: benchmarks, demos, Mythos 5.2 and how to use it</title><link>https://systemdesignsimple.com/claude-fable-5-2-mythos-5-2/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-fable-5-2-mythos-5-2/</guid><description>Claude Fable 5.2 and Mythos 5.2: specs, every benchmark vs Fable 5.1 and Opus 5.5, the OS and Fermat demos, how to use it, and what it changes.</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How can an AI answer with no network? On-device AI explained</title><link>https://systemdesignsimple.com/how-can-ai-answer-with-no-network/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-can-ai-answer-with-no-network/</guid><description>On-device AI answers with no network because the model is files on your disk and a local runtime does the math. Size, memory and speed, worked out.</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Publish/subscribe: why 2 ms jobs waited 31 seconds behind one slow service</title><link>https://systemdesignsimple.com/publish-subscribe-one-slow-service/</link><guid isPermaLink="true">https://systemdesignsimple.com/publish-subscribe-one-slow-service/</guid><description>Publish/subscribe vs direct calls in our simulation: with one 50 ms service among ten, fast jobs averaged 31,484 ms with direct calls and 1.87 ms with pub/sub.</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Service discovery: how 500 services find each other</title><link>https://systemdesignsimple.com/service-discovery/</link><guid isPermaLink="true">https://systemdesignsimple.com/service-discovery/</guid><description>Service discovery in depth: registries, heartbeats, TTLs and leases, client caching, the stale-window formula, and what a faster TTL costs in heartbeat traffic.</description><pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate></item><item><title>AI citation checks: the link opened, but the quote was missing</title><link>https://systemdesignsimple.com/ai-source-quote-wasnt-there/</link><guid isPermaLink="true">https://systemdesignsimple.com/ai-source-quote-wasnt-there/</guid><description>AI citation checks need more than working links. Our synthetic lab shows why exact quote matching catches altered numbers that token overlap accepts.</description><pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to create an AI agent skill: one SKILL.md, reused by two agents</title><link>https://systemdesignsimple.com/how-to-create-an-ai-agent-skill/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-to-create-an-ai-agent-skill/</guid><description>An agent skill is a folder with a SKILL.md file: a name, a one-line description that decides when it loads, and the steps. How we run 23 of them with two agents.</description><pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Multi-tenancy: pool vs silo, tenant context, and noisy neighbor limits</title><link>https://systemdesignsimple.com/one-app-many-restaurants-multi-tenancy/</link><guid isPermaLink="true">https://systemdesignsimple.com/one-app-many-restaurants-multi-tenancy/</guid><description>Multi-tenancy explained: silo vs pool isolation, how a verified tenant ID follows each request, and why shared compute needs per-tenant limits and bounded queues.</description><pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Prompt injection: why the AI obeyed a hidden line</title><link>https://systemdesignsimple.com/why-the-ai-obeyed-a-hidden-line/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-the-ai-obeyed-a-hidden-line/</guid><description>Prompt injection turns untrusted page text into instructions. Learn execution allowlists, argument checks and the limits of prompt-only defences.</description><pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Is Codex CLI still running after you close it? A 5-minute check</title><link>https://systemdesignsimple.com/codex-cli-background-server-check/</link><guid isPermaLink="true">https://systemdesignsimple.com/codex-cli-background-server-check/</guid><description>Since Codex CLI 0.157.0 an interactive session can start a background server. How to see it, how big it is on disk, and the one command that turns it off.</description><pubDate>Sun, 08 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How a leaked password check works without sending the full password</title><link>https://systemdesignsimple.com/is-my-password-leaked-how-the-check-works/</link><guid isPermaLink="true">https://systemdesignsimple.com/is-my-password-leaked-how-the-check-works/</guid><description>How Pwned Passwords checks a SHA-1 prefix, what the lookup reveals, and how to reject breached passwords at sign-up.</description><pubDate>Sun, 08 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Monolith vs microservices: fault isolation and its cost</title><link>https://systemdesignsimple.com/monolith-vs-microservices/</link><guid isPermaLink="true">https://systemdesignsimple.com/monolith-vs-microservices/</guid><description>Monolith vs microservices: worker pools, crash isolation, deployment risks, and the latency and data-consistency costs of splitting services.</description><pubDate>Sun, 08 Nov 2026 00:00:00 GMT</pubDate></item><item><title>AI agent permissions: would an agent delete my files?</title><link>https://systemdesignsimple.com/would-an-ai-agent-delete-my-files/</link><guid isPermaLink="true">https://systemdesignsimple.com/would-an-ai-agent-delete-my-files/</guid><description>AI agent permissions decide what a careless plan can destroy. Full access vs sandbox vs confirm step in one scripted lab: 7 of 10 key files lost, then 0.</description><pubDate>Sun, 08 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to check your Apache OFBiz runs 24.09.08, the release that fixes two pre-login RCEs</title><link>https://systemdesignsimple.com/apache-ofbiz-rce-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/apache-ofbiz-rce-check-and-fix/</guid><description>Every OFBiz before 24.09.08 has CVE-2026-83542 (CVSS 9.8) and CVE-2026-68524, two RCEs without login. Find your real release, upgrade, and verify the service.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How apt proves a package is genuine, and how to check and fix CVE-2026-109518</title><link>https://systemdesignsimple.com/apt-package-signature-flaw-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/apt-package-signature-flaw-check-and-fix/</guid><description>How apt&apos;s signed chain from InRelease to .deb works, how CVE-2026-109518 swapped a package after the check, and which apt versions fix it.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Auto mode is now the default in the Claude apps: what it blocks, how to switch back</title><link>https://systemdesignsimple.com/claude-auto-mode-default/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-auto-mode-default/</guid><description>Claude apps now start in auto mode on Pro, Max and Team. What the automated check blocks, Anthropic&apos;s 89% vs 13.6% test, the limits, and how to go back to manual.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Duet: Claude messages you first, and still asks before it acts</title><link>https://systemdesignsimple.com/claude-duet-proactive-agent/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-duet-proactive-agent/</guid><description>Claude Duet is a proactive agent app: it watches your connected apps, works on its own computer, and asks before it sends, pays, signs or deletes.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude self-assessment: Claude now reports possible misuse in its own conversations</title><link>https://systemdesignsimple.com/claude-self-assessment/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-self-assessment/</guid><description>Claude now writes a hidden safety check after it answers: what self-assessment records, what it does not do to your chat, and the 78% and 94% numbers.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Codex CLI agent command center: what 0.167.0 changes when you run several agents</title><link>https://systemdesignsimple.com/codex-cli-agent-command-center/</link><guid isPermaLink="true">https://systemdesignsimple.com/codex-cli-agent-command-center/</guid><description>Codex CLI 0.167.0 runs the agent command center without the background server (--no-daemon), marks unread results, keeps drafts and previews images.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Did the AI leak my fake password? Secrets in AI prompts</title><link>https://systemdesignsimple.com/did-the-ai-leak-my-fake-password/</link><guid isPermaLink="true">https://systemdesignsimple.com/did-the-ai-leak-my-fake-password/</guid><description>Secrets in AI prompts reach the recipient even when the reply omits them. Lab: a fake password in 40/40 simulated prompts and body logs, 0/40 after redaction.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Server failover: active-passive vs active-active</title><link>https://systemdesignsimple.com/main-server-dies-at-3am/</link><guid isPermaLink="true">https://systemdesignsimple.com/main-server-dies-at-3am/</guid><description>Active-passive vs active-active failover, measured: our simulation lost 111 requests waiting for three 1 s probes and 1 when the first error removed the server.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to check and fix the Node.js Brotli decoder flaw (CVE-2026-111497) in the process that serves traffic</title><link>https://systemdesignsimple.com/nodejs-brotli-rce-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/nodejs-brotli-rce-check-and-fix/</guid><description>Node.js 22.x below 22.23.4, 24.x below 24.22.1 and 26.x below 26.12.1 carry a Brotli decoder memory bug. Upgrade, restart and verify every serving process.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Alpine and musl sshd without PAM: check for the OpenSSH 10.6p2 password bypass and fix it</title><link>https://systemdesignsimple.com/openssh-long-password-auth-bypass-check/</link><guid isPermaLink="true">https://systemdesignsimple.com/openssh-long-password-auth-bypass-check/</guid><description>OpenSSH below 10.6p2 on musl Linux without PAM can accept any long password for accounts stored as *. Check your sshd, upgrade, restart, verify the running build.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to check the PHP build your web server runs and fix CVE-2026-109618 and CVE-2026-109619</title><link>https://systemdesignsimple.com/php-rce-php2shell-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/php-rce-php2shell-check-and-fix/</guid><description>PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35 fix two CVSS 9.2 flaws in PHP-FPM, mod_php and php-cgi. Check the build serving traffic, update, restart, verify.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>PHP2Shell and Node2Shell: why they were so dangerous, and how to patch your servers</title><link>https://systemdesignsimple.com/php2shell-node2shell-why-dangerous-how-to-patch/</link><guid isPermaLink="true">https://systemdesignsimple.com/php2shell-node2shell-why-dangerous-how-to-patch/</guid><description>Two remote code execution bugs fixed on the same day, exploited within a week. How the fixes were prepared, why they matter, and how to patch the right process.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to check Rails image uploads and lock down ImageMagick (GHSA-f3w6-w445-fwh9)</title><link>https://systemdesignsimple.com/rails-imagemagick-upload-flaw-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/rails-imagemagick-upload-flaw-check-and-fix/</guid><description>Rails Active Storage with ImageMagick: how uploads reach ImageMagick coders, which versions fix it, and how to lock down policy.xml. Our lab checked 9 files.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to check WordPress for CVE-2026-119977: the version, the auto-update blockers and the upgrade folder</title><link>https://systemdesignsimple.com/wordpress-core-upgrade-flaw-check-and-fix/</link><guid isPermaLink="true">https://systemdesignsimple.com/wordpress-core-upgrade-flaw-check-and-fix/</guid><description>Check if your WordPress core is below 7.1.4 or its branch fix for CVE-2026-119977, and find the setting, plugin or deploy that stops security releases installing.</description><pubDate>Sat, 07 Nov 2026 00:00:00 GMT</pubDate></item><item><title>1,000 people ask AI at once. Who gets an answer first? KV cache and continuous batching</title><link>https://systemdesignsimple.com/1000-people-ask-ai-at-once/</link><guid isPermaLink="true">https://systemdesignsimple.com/1000-people-ask-ai-at-once/</guid><description>How one AI server answers 1,000 people at once: the KV cache and continuous batching. Our toy-model lab: 567 ms to 39 ms, median first-token wait 452 s to 14 s.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>I interrupted the AI. Why did it keep talking? Voice AI barge-in explained</title><link>https://systemdesignsimple.com/i-interrupted-the-ai-why-did-it-keep-talking/</link><guid isPermaLink="true">https://systemdesignsimple.com/i-interrupted-the-ai-why-did-it-keep-talking/</guid><description>Voice AI barge-in explained: our simulation cut talk-over from 4,075 ms to 155 ms, but the energy detector also stopped on 19 of 20 synthetic coughs.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Login anomaly detection: why our 3 a.m. alarm was mostly wrong</title><link>https://systemdesignsimple.com/login-anomaly-detection-3am-rule/</link><guid isPermaLink="true">https://systemdesignsimple.com/login-anomaly-detection-3am-rule/</guid><description>Login anomaly detection: our synthetic lab compared a fixed night rule with per-user baselines, cutting false alarms from 667 to 32 across 8,457 logins.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Move 1 button: why a computer-use agent checks after every click</title><link>https://systemdesignsimple.com/move-one-button-can-ai-finish-the-task/</link><guid isPermaLink="true">https://systemdesignsimple.com/move-one-button-can-ai-finish-the-task/</guid><description>How a computer-use agent copes when a button moves: in 50 simulated runs, replay finished 2, looking first 43, and checking after each click 47.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Node.js security: 4 common vulnerabilities we reproduced and fixed</title><link>https://systemdesignsimple.com/nodejs-security-common-vulnerabilities/</link><guid isPermaLink="true">https://systemdesignsimple.com/nodejs-security-common-vulnerabilities/</guid><description>SQL by template string, path.join on user input, a deep merge that walks into __proto__, and exec with a shell string: reproduced in our lab, then fixed.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Python security: 4 common vulnerabilities we reproduced and fixed</title><link>https://systemdesignsimple.com/python-security-common-vulnerabilities/</link><guid isPermaLink="true">https://systemdesignsimple.com/python-security-common-vulnerabilities/</guid><description>SQL in an f-string, pickle.loads on client bytes, subprocess with shell=True and open(base / name): reproduced in our Python 3.14.7 lab, then fixed.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Same question 10 times, different answers. Which one passes? AI evals and guardrails</title><link>https://systemdesignsimple.com/same-question-ten-answers-which-passes/</link><guid isPermaLink="true">https://systemdesignsimple.com/same-question-ten-answers-which-passes/</guid><description>AI evals and guardrails check replies before delivery. Our seeded simulation shows what format, refund-math and phrase checks catch, and what they miss.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>SQL vs NoSQL: choose the data shape first</title><link>https://systemdesignsimple.com/sql-or-nosql-wrong-question/</link><guid isPermaLink="true">https://systemdesignsimple.com/sql-or-nosql-wrong-question/</guid><description>SQL vs NoSQL is the wrong first question. Our lab: renaming one product rewrote 47,999 embedded documents in about 630 ms, while normalized tables changed 1 row.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Semantic cache: can AI safely reuse a similar answer?</title><link>https://systemdesignsimple.com/two-questions-look-alike-can-ai-reuse-the-answer/</link><guid isPermaLink="true">https://systemdesignsimple.com/two-questions-look-alike-can-ai-reuse-the-answer/</guid><description>A semantic cache reuses a saved AI answer for a similar question. In our lexical 60-pair lab, every cutoff we tested still made wrong reuses.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Vertical vs horizontal scaling: one server or eight?</title><link>https://systemdesignsimple.com/vertical-vs-horizontal-scaling/</link><guid isPermaLink="true">https://systemdesignsimple.com/vertical-vs-horizontal-scaling/</guid><description>Vertical vs horizontal scaling in our simulation: one 8-core server vs eight 1-core servers, a near tie when matched, 30.23% vs 0.54% failed after a crash.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>What is SQL injection and how to prevent it: bound parameters, tested in a PHP lab</title><link>https://systemdesignsimple.com/what-is-sql-injection-and-how-to-prevent-it/</link><guid isPermaLink="true">https://systemdesignsimple.com/what-is-sql-injection-and-how-to-prevent-it/</guid><description>SQL injection happens when input becomes part of the SQL text. Bound parameters fix it. Our PHP lab: 3 rows with string-built SQL, 0 with prepared statements.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>&quot;Hello&quot; went to a small AI, maths went to a big one. Who decided? Model routing in an LLM gateway</title><link>https://systemdesignsimple.com/who-decides-which-ai-answers-your-message/</link><guid isPermaLink="true">https://systemdesignsimple.com/who-decides-which-ai-answers-your-message/</guid><description>Model routing lets an LLM gateway choose which model answers. Our 60-prompt lab shows how keyword rules affect routing and illustrative costs.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Why does AI say &quot;too many requests&quot; after one message?</title><link>https://systemdesignsimple.com/why-ai-says-too-many-requests-after-one-message/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-ai-says-too-many-requests-after-one-message/</guid><description>Why AI says too many requests after one message: token budgets, shared limits, and a reproducible lab showing rejection with request credits remaining.</description><pubDate>Fri, 06 Nov 2026 00:00:00 GMT</pubDate></item><item><title>The cable to Asia breaks. Your bank must choose: correct balance or an answer</title><link>https://systemdesignsimple.com/cable-to-asia-breaks/</link><guid isPermaLink="true">https://systemdesignsimple.com/cable-to-asia-breaks/</guid><description>CAP theorem with real lab numbers: in our simulated 20 s cable cut, CP refused 50.7% of requests with 0 mismatched reads. AP answered 100%, but 241 reads mismatched.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>RAG retrieval: how AI finds one line in 100 pages</title><link>https://systemdesignsimple.com/how-ai-finds-one-line-in-100-pages/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-ai-finds-one-line-in-100-pages/</guid><description>RAG retrieval ranks document chunks before generation. Explore BM25 internals, hybrid search, token budgets and a reproducible 100-page lab.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>MCP tool discovery: 3 new apps, 0 changes to the client code</title><link>https://systemdesignsimple.com/how-ai-learns-new-apps-tool-discovery/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-ai-learns-new-apps-tool-discovery/</guid><description>How tool discovery lets one AI client use new apps unchanged: the tools/list handshake, a lab with 3 apps, 55 of 60 right picks, and the vague-description drop.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>HNSW vector search: how our lab skipped 96% of 20,000 synthetic vectors</title><link>https://systemdesignsimple.com/how-search-skips-96-percent-of-vectors/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-search-skips-96-percent-of-vectors/</guid><description>HNSW vector search with a lab: 20,000 vectors, 776 touched per query, recall 0.944, the efSearch dial, the run-to-run speed spread, and the B-tree contrast.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Layer 4 vs Layer 7 Load Balancer: Internals and Trade-offs</title><link>https://systemdesignsimple.com/layer-4-vs-layer-7-load-balancer/</link><guid isPermaLink="true">https://systemdesignsimple.com/layer-4-vs-layer-7-load-balancer/</guid><description>Layer 4 vs layer 7 load balancer internals: connection tracking, HTTP routing, keep-alive imbalance, draining, and measured lab costs.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Eventual consistency: stale likes and safe withdrawals</title><link>https://systemdesignsimple.com/like-count-wrong-balance-never/</link><guid isPermaLink="true">https://systemdesignsimple.com/like-count-wrong-balance-never/</guid><description>Eventual consistency left 59.2% of lab reads stale. Learn why quorum reads can go backwards and atomic conditional debits prevent overdrafts.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Load balancer vs reverse proxy: one door, two jobs</title><link>https://systemdesignsimple.com/load-balancer-vs-reverse-proxy/</link><guid isPermaLink="true">https://systemdesignsimple.com/load-balancer-vs-reverse-proxy/</guid><description>Load balancer vs reverse proxy: connection handling, Nginx algorithms, caching, failure modes, and verified simulation results.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Context window: why a long AI chat forgets your name</title><link>https://systemdesignsimple.com/my-chat-forgot-my-name/</link><guid isPermaLink="true">https://systemdesignsimple.com/my-chat-forgot-my-name/</guid><description>How a context window limits chat history: a reproducible simulation of truncation, pinned memory, rolling summaries, and their engineering trade-offs.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>One database writes, ten read: primary and read replicas</title><link>https://systemdesignsimple.com/one-database-writes-ten-read/</link><guid isPermaLink="true">https://systemdesignsimple.com/one-database-writes-ten-read/</guid><description>One primary writes, read replicas serve reads. In our simulation 10 replicas gave 10.3x the reads, but at 500 ms lag your own post was missing 51.8% of the time.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Series vs parallel availability: two servers in a row</title><link>https://systemdesignsimple.com/two-servers-in-a-row/</link><guid isPermaLink="true">https://systemdesignsimple.com/two-servers-in-a-row/</guid><description>Series vs parallel availability: two independent 99% servers give 98.01% when both are required, or 99.99% when either copy can serve the request.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>When will AGI happen? Why every forecast names a different finish line</title><link>https://systemdesignsimple.com/when-will-agi-happen/</link><guid isPermaLink="true">https://systemdesignsimple.com/when-will-agi-happen/</guid><description>When will AGI happen? What lab leaders, forecasters and a researcher survey said from 2023 to 2026, each date with its definition, and what METR measures.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Load balancing algorithms: which server gets your request?</title><link>https://systemdesignsimple.com/which-server-gets-your-request/</link><guid isPermaLink="true">https://systemdesignsimple.com/which-server-gets-your-request/</guid><description>Load balancing algorithms: round robin, least connections, user-ID hash. In our 4-server simulation, quick-request p99 was 150.3 ms, 22 ms and 468.9 ms.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>AI agent tokens: why history makes total input grow</title><link>https://systemdesignsimple.com/why-an-ai-agent-re-sends-its-history/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-an-ai-agent-re-sends-its-history/</guid><description>AI agent tokens add up: each call carries the retained history as input, so an append-only task grows close to N squared. The math, KV caching and trimming.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Embeddings explained: why a search for car finds vehicle</title><link>https://systemdesignsimple.com/why-search-for-car-finds-vehicle/</link><guid isPermaLink="true">https://systemdesignsimple.com/why-search-for-car-finds-vehicle/</guid><description>Embeddings explained with BM25, PPMI and cosine arithmetic: why car finds vehicle, where retrieval fails, and how hybrid vector search scales.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>What happens when you type google.com? DNS in depth</title><link>https://systemdesignsimple.com/you-type-google-com-100-ms/</link><guid isPermaLink="true">https://systemdesignsimple.com/you-type-google-com-100-ms/</guid><description>What happens when you type google.com? DNS walks root, .com and name server, then caches. Our model: 37 ms cold, 7.1 ms resolver hit, 0.05 ms OS cache.</description><pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate></item><item><title>One AI writes, another reviews: designing a multi-agent coding pipeline</title><link>https://systemdesignsimple.com/one-ai-writes-one-ai-reviews/</link><guid isPermaLink="true">https://systemdesignsimple.com/one-ai-writes-one-ai-reviews/</guid><description>How to split AI coding work across agents: routing, a second model as reviewer, a gate with a round cap, and a sandbox. With our own review data.</description><pubDate>Wed, 04 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How to use Claude like a professional, not like a casual user</title><link>https://systemdesignsimple.com/claude-like-a-pro/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-like-a-pro/</guid><description>Use Claude like a pro: Projects, Skills, memory and connectors in the apps; CLAUDE.md, hooks, permissions and headless runs in Claude Code; habits that catch errors.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Ultra: the 5-hour limit is gone, the weekly cap stays</title><link>https://systemdesignsimple.com/claude-ultra-no-5-hour-limit/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-ultra-no-5-hour-limit/</guid><description>Claude Ultra plan limits explained: no 5-hour limit for verified accounts, a weekly cap of 10x Pro, what the old Max 20x name measured, and how verification works.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Fast for you, slow for a million. Why?</title><link>https://systemdesignsimple.com/fast-alone-slow-crowded/</link><guid isPermaLink="true">https://systemdesignsimple.com/fast-alone-slow-crowded/</guid><description>A 5 ms server against a 15 ms server: why the faster one has a 498 ms median at 100 users, explained with queueing laws, pools and a simulation.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>A highway with 10 lanes can still be slow</title><link>https://systemdesignsimple.com/ten-lanes-still-slow/</link><guid isPermaLink="true">https://systemdesignsimple.com/ten-lanes-still-slow/</guid><description>More workers raise throughput, not speed: 1 to 10 lanes took a simulation from 5 to 49.8 req/s at 200 ms, until one 50 ms lock capped it at 19.9 req/s.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>50K fans, 5K seats. Design it.</title><link>https://systemdesignsimple.com/50k-fans-5k-seats/</link><guid isPermaLink="true">https://systemdesignsimple.com/50k-fans-5k-seats/</guid><description>Performance vs scalability for engineers: Little&apos;s law, utilisation vs queueing delay, Amdahl and the USL, load balancing, retry storms and admission control.</description><pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Design Instagram&apos;s feed: system design in five steps</title><link>https://systemdesignsimple.com/design-instagram-feed/</link><guid isPermaLink="true">https://systemdesignsimple.com/design-instagram-feed/</guid><description>System design as engineering, on a news feed: API, data model, fan-out on read vs write with cost formulas, Redis timelines, cursors, hot keys and a benchmark.</description><pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate></item><item><title>1M rides a day. What do you do in the first 5 minutes?</title><link>https://systemdesignsimple.com/ride-app-requirements/</link><guid isPermaLink="true">https://systemdesignsimple.com/ride-app-requirements/</guid><description>Requirements in depth: functional vs non-functional, SLIs, SLOs and error budgets, nines arithmetic, capacity estimation and how the numbers pick the design.</description><pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Two fans click the same seat at the same moment. Who wins?</title><link>https://systemdesignsimple.com/two-fans-one-seat/</link><guid isPermaLink="true">https://systemdesignsimple.com/two-fans-one-seat/</guid><description>Race conditions, lost updates and write skew, how PostgreSQL isolation levels, row locks and conditional updates prevent them, and when strong consistency matters.</description><pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude Code vs Codex CLI: how we use both AI coding agents together</title><link>https://systemdesignsimple.com/claude-code-vs-codex-cli/</link><guid isPermaLink="true">https://systemdesignsimple.com/claude-code-vs-codex-cli/</guid><description>Hands-on comparison of Claude Code and Codex CLI from daily work, and why letting one review the other catches more mistakes.</description><pubDate>Sun, 01 Nov 2026 00:00:00 GMT</pubDate></item><item><title>You tap like. How does your friend&apos;s phone know a second later?</title><link>https://systemdesignsimple.com/how-a-like-reaches-your-friend/</link><guid isPermaLink="true">https://systemdesignsimple.com/how-a-like-reaches-your-friend/</guid><description>From your tap to your friend&apos;s buzz: events, a queue, a worker, live connections, Apple and Google push, and grouping 1,000 likes into one alert.</description><pubDate>Sun, 01 Nov 2026 00:00:00 GMT</pubDate></item><item><title>How do apps handle a million likes at once without crashing?</title><link>https://systemdesignsimple.com/million-likes-at-once/</link><guid isPermaLink="true">https://systemdesignsimple.com/million-likes-at-once/</guid><description>When millions tap like at once, one database row becomes a traffic jam. How big apps survive: a queue, split counters and a cache.</description><pubDate>Sun, 01 Nov 2026 00:00:00 GMT</pubDate></item><item><title>PHP security: 7 common vulnerabilities and how to fix them</title><link>https://systemdesignsimple.com/php-security-common-vulnerabilities/</link><guid isPermaLink="true">https://systemdesignsimple.com/php-security-common-vulnerabilities/</guid><description>Defensive PHP checklist: SQL injection, XSS, CSRF, password storage, uploads, sessions and unserialize, with the exact fix for each.</description><pubDate>Sun, 01 Nov 2026 00:00:00 GMT</pubDate></item></channel></rss>