https://binh-ai.com/en/binhAI collects AI tips, tricks and the best reads on AI for developers, PMs and everyday users.enhttps://binh-ai.com/en/posts/memory-safety-c-cpp-70-percenthttps://binh-ai.com/en/posts/memory-safety-c-cpp-70-percentFri, 30 Oct 2026 00:00:00 +0000System DesignData from Microsoft, Chromium and Android, the common memory bugs (buffer overflows, use-after-free…) and why careful coding or sandboxing isn't enough.https://binh-ai.com/en/posts/rust-memory-safety-how-it-workshttps://binh-ai.com/en/posts/rust-memory-safety-how-it-worksFri, 30 Oct 2026 00:00:00 +0000System DesignOwnership, borrowing and lifetimes; other safeguards; Android data (~1000× lower memory vulnerability density) and the limits of unsafe Rust.https://binh-ai.com/en/posts/migrate-c-cpp-to-rust-strategyhttps://binh-ai.com/en/posts/migrate-c-cpp-to-rust-strategyFri, 30 Oct 2026 00:00:00 +0000AI AutomationAndroid's lesson: vulnerabilities cluster in new code, so write new code in Rust; a 4-step roadmap, the CISA/NSA framework, cxx/FFI and AI translation.https://binh-ai.com/en/posts/claude-code-prompt-cache-what-breaks-ithttps://binh-ai.com/en/posts/claude-code-prompt-cache-what-breaks-itFri, 30 Oct 2026 00:00:00 +0000Claude CodeSwitching model, effort, fast mode, MCP servers or /compact breaks the cache; editing files, modes and rewinding don't. Plus 5-minute/1-hour TTLs and /usage.https://binh-ai.com/en/posts/claude-plugin-eval-test-skillshttps://binh-ai.com/en/posts/claude-plugin-eval-test-skillsFri, 30 Oct 2026 00:00:00 +0000AI Agents & SkillsThe claude plugin eval command runs skills on real cases against a no-plugin baseline: six grader types, tips for stable cases and costs to know.https://binh-ai.com/en/posts/claude-api-compaction-long-agentshttps://binh-ai.com/en/posts/claude-api-compaction-long-agentsFri, 30 Oct 2026 00:00:00 +0000Claude APIServer-side compaction at a token threshold or on demand: parameters, the compaction block, counting usage.iterations, prompt caching and risks.https://binh-ai.com/en/posts/claude-code-auto-memoryhttps://binh-ai.com/en/posts/claude-code-auto-memoryFri, 30 Oct 2026 00:00:00 +0000Claude CodeHow auto memory differs from CLAUDE.md, the four note types, MEMORY.md's 200-line/25KB load limit, toggling it, relocating it, and token and security caveats.https://binh-ai.com/en/posts/claude-api-agent-skillshttps://binh-ai.com/en/posts/claude-api-agent-skillsFri, 30 Oct 2026 00:00:00 +0000Claude APIAttaching skills in the Messages API: code execution tool, container.skills, pre-built pptx/xlsx/docx/pdf skills, /v1/skills uploads, output files.https://binh-ai.com/en/posts/claude-code-output-styleshttps://binh-ai.com/en/posts/claude-code-output-stylesThu, 29 Oct 2026 00:00:00 +0000Claude CodeThe four built-in styles (Proactive, Concise, Explanatory, Learning), switching, custom styles, and styles vs CLAUDE.md, skills, hooks.https://binh-ai.com/en/posts/claude-code-permission-ruleshttps://binh-ai.com/en/posts/claude-code-permission-rulesThu, 29 Oct 2026 00:00:00 +0000Claude CodePermission modes, rule syntax for Bash, Read, WebFetch and MCP, the deny → ask → allow order, and the limits of Bash rules you need to know.https://binh-ai.com/en/posts/claude-code-mcp-scopeshttps://binh-ai.com/en/posts/claude-code-mcp-scopesThu, 29 Oct 2026 00:00:00 +0000Claude CodeAdding stdio/HTTP MCP servers, the three scopes and precedence, sharing .mcp.json without secrets, approving project servers.https://binh-ai.com/en/posts/claude-api-streaming-ssehttps://binh-ai.com/en/posts/claude-api-streaming-sseThu, 29 Oct 2026 00:00:00 +0000Claude APIEnabling streaming, SDK helpers, the SSE event order, delta types (text, tool JSON, thinking) and handling overloaded errors mid-stream.https://binh-ai.com/en/posts/claude-api-citationshttps://binh-ai.com/en/posts/claude-api-citationsThu, 29 Oct 2026 00:00:00 +0000Claude APIEnabling citations for text, PDF and custom documents: citation formats, token costs, versus prompt-based citations, and the structured outputs limitation.https://binh-ai.com/en/posts/claude-api-web-search-toolhttps://binh-ai.com/en/posts/claude-api-web-search-toolThu, 29 Oct 2026 00:00:00 +0000Claude APIServer-side web search: tool versions, dynamic filtering, max_uses, domain filters, $10 per 1,000 searches, errors and injection risk.https://binh-ai.com/en/posts/scaling-laws-chinchillahttps://binh-ai.com/en/posts/scaling-laws-chinchillaThu, 29 Oct 2026 00:00:00 +0000LLM & NLPFrom Kaplan's 2020 power laws to Chinchilla's ~20 tokens per parameter, and why models are now over-trained to be cheap to serve.https://binh-ai.com/en/posts/rlhf-constitutional-aihttps://binh-ai.com/en/posts/rlhf-constitutional-aiThu, 29 Oct 2026 00:00:00 +0000LLM & NLPInstructGPT's three RLHF steps, how Constitutional AI uses principles and AI feedback, a comparison of the two, and side effects like sycophancy and verbosity.https://binh-ai.com/en/posts/chain-of-thought-reasoninghttps://binh-ai.com/en/posts/chain-of-thought-reasoningThu, 29 Oct 2026 00:00:00 +0000LLM & NLPThree foundational studies (CoT, step-by-step, self-consistency), why step-by-step reasoning works, and using it with Claude's thinking.https://binh-ai.com/en/posts/mixture-of-experts-moehttps://binh-ai.com/en/posts/mixture-of-experts-moeThu, 29 Oct 2026 00:00:00 +0000LLM & NLPRouters and experts, three milestones (Sparsely-Gated MoE, Switch Transformer, Mixtral), and understanding total vs active parameters when self-hosting.https://binh-ai.com/en/posts/react-toolformer-tool-use-originshttps://binh-ai.com/en/posts/react-toolformer-tool-use-originsThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsThe papers behind tool-using agents: ReAct interleaves reasoning and acting, Toolformer learns when to call APIs; plus inherited risks.https://binh-ai.com/en/posts/multi-agent-research-system-lessonshttps://binh-ai.com/en/posts/multi-agent-research-system-lessonsThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsThe orchestrator–worker architecture behind Research, performance and token cost figures, prompting principles, evaluation and when not to use multiple agents.https://binh-ai.com/en/posts/long-running-agent-harnesshttps://binh-ai.com/en/posts/long-running-agent-harnessThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsHow Anthropic uses an initializer agent, coding agent, JSON feature list, progress file, init.sh and Git to run multi-day agent projects.https://binh-ai.com/en/posts/prompt-chaining-self-correctionhttps://binh-ai.com/en/posts/prompt-chaining-self-correctionThu, 29 Oct 2026 00:00:00 +0000TipsWhen to split work into a chain of API calls, the draft–review–refine self-correction pattern, chain design principles and the cost and latency risks.https://binh-ai.com/en/posts/migrate-away-from-prefillhttps://binh-ai.com/en/posts/migrate-away-from-prefillThu, 29 Oct 2026 00:00:00 +0000TipsFrom Claude 4.6, last-turn prefill returns a 400. Migration by purpose: JSON, classification, preambles, continuations, context reminders.https://binh-ai.com/en/posts/llm-as-a-judgehttps://binh-ai.com/en/posts/llm-as-a-judgeThu, 29 Oct 2026 00:00:00 +0000TipsMT-Bench found LLM judges agree with humans over 80% of the time. Three grading modes, four biases and a sample judge prompt.https://binh-ai.com/en/posts/agentic-coding-prompt-pitfallshttps://binh-ai.com/en/posts/agentic-coding-prompt-pitfallsThu, 29 Oct 2026 00:00:00 +0000TipsOver-engineering, coding to the tests, leftover temp files, guessing at unread code: four coding-agent habits and Anthropic's sample fixes.https://binh-ai.com/en/posts/ai-developer-productivity-researchhttps://binh-ai.com/en/posts/ai-developer-productivity-researchThu, 29 Oct 2026 00:00:00 +0000AI AutomationCopilot's experiment (55.8% faster) vs METR 2025 (experienced developers 19% slower): why they differ and how to measure AI productivity.https://binh-ai.com/en/posts/human-ai-interaction-18-guidelineshttps://binh-ai.com/en/posts/human-ai-interaction-18-guidelinesThu, 29 Oct 2026 00:00:00 +0000AI AutomationA checklist of 18 guidelines in 4 phases (initially, during interaction, when wrong, over time) and how to apply them to today's chatbots, copilots and agents.https://binh-ai.com/en/posts/rules-of-ml-for-llm-featureshttps://binh-ai.com/en/posts/rules-of-ml-for-llm-featuresThu, 29 Oct 2026 00:00:00 +0000AI AutomationThe key rules from Google's Rules of Machine Learning and a four-step path (baseline, minimal version, shadow, gradual launch) for LLM features.https://binh-ai.com/en/posts/ollama-run-llm-locallyhttps://binh-ai.com/en/posts/ollama-run-llm-locallyThu, 29 Oct 2026 00:00:00 +0000ToolsInstall and run open models with one command, call the REST API on port 11434, customise with a Modelfile, and decide between local and cloud APIs.https://binh-ai.com/en/posts/vllm-pagedattention-servinghttps://binh-ai.com/en/posts/vllm-pagedattention-servingThu, 29 Oct 2026 00:00:00 +0000ToolsWhy the KV cache wastes GPU memory, how PagedAttention pages it, continuous batching, an OpenAI-compatible server, and vLLM vs Ollama.https://binh-ai.com/en/posts/markitdown-files-to-markdownhttps://binh-ai.com/en/posts/markitdown-files-to-markdownThu, 29 Oct 2026 00:00:00 +0000ToolsMicrosoft's Python library that converts most file types to structured Markdown: CLI and Python usage, fitting it into a RAG pipeline, and security warnings.https://binh-ai.com/en/posts/owasp-top-10-llm-2025https://binh-ai.com/en/posts/owasp-top-10-llm-2025Thu, 29 Oct 2026 00:00:00 +0000System DesignThe ten most common security risks in LLM applications per OWASP 2025, from prompt injection to unbounded consumption, with core mitigations for each.https://binh-ai.com/en/posts/api-rate-limiting-token-buckethttps://binh-ai.com/en/posts/api-rate-limiting-token-bucketThu, 29 Oct 2026 00:00:00 +0000System DesignThe token bucket algorithm, the four limiters Stripe runs in production, and applying them to AI apps: token-based limits, concurrency, dark launches.https://binh-ai.com/en/posts/idempotency-keys-safe-retrieshttps://binh-ai.com/en/posts/idempotency-keys-safe-retriesThu, 29 Oct 2026 00:00:00 +0000System DesignStripe's idempotency keys and the IETF Idempotency-Key header, the cases servers must handle, pairing with backoff retries, and applying it to AI agent tools.https://binh-ai.com/en/posts/bitter-lesson-suttonhttps://binh-ai.com/en/posts/bitter-lesson-suttonThu, 29 Oct 2026 00:00:00 +0000ReadingRichard Sutton's 2019 essay summarised: general methods that leverage compute win, four examples, the link to LLMs and lessons for products.https://binh-ai.com/en/posts/software-2-0-karpathyhttps://binh-ai.com/en/posts/software-2-0-karpathyThu, 29 Oct 2026 00:00:00 +0000ReadingAndrej Karpathy's 2017 essay summarised: Software 1.0 vs 2.0, benefits, limits, and why data, evals and prompts are source code in the LLM era.https://binh-ai.com/en/posts/chip-huyen-genai-platformhttps://binh-ai.com/en/posts/chip-huyen-genai-platformThu, 29 Oct 2026 00:00:00 +0000ReadingA summary of Building A Generative AI Platform: context, guardrails, router and gateway, caching, write actions, plus observability and orchestration.https://binh-ai.com/en/posts/hamel-your-ai-product-needs-evalshttps://binh-ai.com/en/posts/hamel-your-ai-product-needs-evalsThu, 29 Oct 2026 00:00:00 +0000ReadingHamel Husain's post summarised: the Lucy assistant story, three eval levels (assertions, human and LLM judges, A/B tests) and reading real data.https://binh-ai.com/en/posts/llm-tokenization-bpehttps://binh-ai.com/en/posts/llm-tokenization-bpeThu, 29 Oct 2026 00:00:00 +0000LLM & NLPTokenization and Byte Pair Encoding explained: why models miscount letters, why non-English text costs more tokens, and how to count tokens with the Claude API.https://binh-ai.com/en/posts/transformer-self-attention-explainedhttps://binh-ai.com/en/posts/transformer-self-attention-explainedThu, 29 Oct 2026 00:00:00 +0000LLM & NLPSelf-attention, Query/Key/Value and multi-head explained, plus why the Transformer makes long context expensive and caching effective.https://binh-ai.com/en/posts/llm-sampling-temperature-top-phttps://binh-ai.com/en/posts/llm-sampling-temperature-top-pThu, 29 Oct 2026 00:00:00 +0000LLM & NLPHow sampling works in LLMs, what temperature, top-p and top-k do, how to set them on the Claude API per task, and common misconceptions.https://binh-ai.com/en/posts/embeddings-semantic-searchhttps://binh-ai.com/en/posts/embeddings-semantic-searchThu, 29 Oct 2026 00:00:00 +0000LLM & NLPWhat embeddings are, why meaning-based search beats keywords, using them with Claude via Voyage AI, input_type, quantization and common traps.https://binh-ai.com/en/posts/contextual-retrieval-raghttps://binh-ai.com/en/posts/contextual-retrieval-ragThu, 29 Oct 2026 00:00:00 +0000LLM & NLPAnthropic's Contextual Retrieval: prepend context to each chunk and combine embeddings, BM25 and reranking to cut retrieval failures by 67%.https://binh-ai.com/en/posts/why-llms-hallucinatehttps://binh-ai.com/en/posts/why-llms-hallucinateThu, 29 Oct 2026 00:00:00 +0000LLM & NLPThe causes of hallucination per 2025 research (training plus scoring that rewards guessing), common types, and Anthropic's mitigation techniques.https://binh-ai.com/en/posts/context-rot-lost-in-the-middlehttps://binh-ai.com/en/posts/context-rot-lost-in-the-middleThu, 29 Oct 2026 00:00:00 +0000LLM & NLPTwo studies show LLMs use context unevenly: information in the middle gets missed and quality drops as input grows. Plus six mitigation tactics.https://binh-ai.com/en/posts/mcp-2026-stateless-spechttps://binh-ai.com/en/posts/mcp-2026-stateless-specThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsThe major changes in the MCP 2026-07-28 spec: no handshake or sessions, server/discover, subscriptions/listen, MRTR, deprecations and how to upgrade.https://binh-ai.com/en/posts/build-mcp-server-pythonhttps://binh-ai.com/en/posts/build-mcp-server-pythonThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsWrite an MCP server with the official Python SDK, test it with MCP Inspector, connect Claude Desktop and Claude Code, and fix common errors.https://binh-ai.com/en/posts/mcp-remote-server-oauth-securityhttps://binh-ai.com/en/posts/mcp-remote-server-oauth-securityThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsPutting an MCP server on the internet safely: Origin checks, the OAuth 2.1 flow with PKCE and resource, client registration via CIMD, no token passthrough.https://binh-ai.com/en/posts/claude-code-remote-control-phonehttps://binh-ai.com/en/posts/claude-code-remote-control-phoneThu, 29 Oct 2026 00:00:00 +0000Claude CodeContinue a local Claude Code session from the Claude app or a browser: three ways to start, server mode, worktrees, no inbound ports.https://binh-ai.com/en/posts/claude-code-chrome-browser-testinghttps://binh-ai.com/en/posts/claude-code-chrome-browser-testingThu, 29 Oct 2026 00:00:00 +0000Claude CodeConnect Claude Code to Chrome via the Claude in Chrome extension: console debugging, form testing, data extraction, site permissions and risks.https://binh-ai.com/en/posts/claude-code-cloud-scheduled-taskshttps://binh-ai.com/en/posts/claude-code-cloud-scheduled-tasksThu, 29 Oct 2026 00:00:00 +0000AI AutomationCreate cloud tasks with /schedule: schedule, API and GitHub triggers, the run environment, connectors, UTC cron and caveats.https://binh-ai.com/en/posts/claude-code-auto-mode-classifierhttps://binh-ai.com/en/posts/claude-code-auto-mode-classifierThu, 29 Oct 2026 00:00:00 +0000Claude CodeAuto mode uses a classifier to review each action: default blocks, how to enable it, autoMode settings, cost, and when you still need deny rules.https://binh-ai.com/en/posts/claude-code-sandbox-bashhttps://binh-ai.com/en/posts/claude-code-sandbox-bashThu, 29 Oct 2026 00:00:00 +0000Claude CodeTurn on the sandboxed Bash tool: Seatbelt, bubblewrap, allowedDomains, denyRead, excludedCommands and pitfalls to avoid.https://binh-ai.com/en/posts/claude-code-review-github-prhttps://binh-ai.com/en/posts/claude-code-review-github-prThu, 29 Oct 2026 00:00:00 +0000Claude CodeClaude's managed PR review for Team/Enterprise: find-verify-rank-post, Quick/Standard/Deep levels, REVIEW.md, and how it compares with other review options.https://binh-ai.com/en/posts/claude-code-dynamic-workflows-ultracodehttps://binh-ai.com/en/posts/claude-code-dynamic-workflows-ultracodeThu, 29 Oct 2026 00:00:00 +0000Claude CodeWorkflows are JavaScript scripts orchestrating subagents at scale: /deep-research, the ultracode keyword, /workflows, limits and cost.https://binh-ai.com/en/posts/claude-code-advisor-toolhttps://binh-ai.com/en/posts/claude-code-advisor-toolThu, 29 Oct 2026 00:00:00 +0000Claude CodePair a fast main model with a stronger advisor in Claude Code: enabling /advisor, model pairings, cost, limits and how it compares with opusplan.https://binh-ai.com/en/posts/claude-code-computer-use-clihttps://binh-ai.com/en/posts/claude-code-computer-use-cliThu, 29 Oct 2026 00:00:00 +0000Claude CodeEnable computer-use via /mcp so Claude can test native apps, simulators and layout bugs: macOS permissions, per-app approval, safeguards and risks.https://binh-ai.com/en/posts/claude-code-channels-telegramhttps://binh-ai.com/en/posts/claude-code-channels-telegramThu, 29 Oct 2026 00:00:00 +0000AI AutomationPush chat messages and webhooks into a local Claude Code session with channels: Telegram, Discord and iMessage setup, allowlists and security risks.https://binh-ai.com/en/posts/claude-code-statusline-context-costhttps://binh-ai.com/en/posts/claude-code-statusline-context-costThu, 29 Oct 2026 00:00:00 +0000Claude CodeSet up a Claude Code status line with /statusline or a jq script to show context %, session cost, 5-hour limit and effort, and save tokens.https://binh-ai.com/en/posts/claude-code-checkpoints-rewindhttps://binh-ai.com/en/posts/claude-code-checkpoints-rewindThu, 29 Oct 2026 00:00:00 +0000Claude CodeAutomatic checkpoints before every prompt; /rewind restores code, conversation, or summarizes to free context. What checkpoints don't track.https://binh-ai.com/en/posts/claude-code-worktrees-parallel-sessionshttps://binh-ai.com/en/posts/claude-code-worktrees-parallel-sessionsThu, 29 Oct 2026 00:00:00 +0000Claude Codeclaude --worktree gives each session its own checkout; .worktreeinclude copies .env files; subagent isolation: worktree; cleanup rules.https://binh-ai.com/en/posts/claude-code-plugins-marketplacehttps://binh-ai.com/en/posts/claude-code-plugins-marketplaceThu, 29 Oct 2026 00:00:00 +0000Claude CodePlugins bundle skills, subagents, hooks and MCP into one /plugin install. Install scopes, marketplaces, per-turn context cost and security.https://binh-ai.com/en/posts/skill-authoring-best-practiceshttps://binh-ai.com/en/posts/skill-authoring-best-practicesThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsSKILL.md principles: be concise, pick degrees of freedom, name and describe for discovery, progressive disclosure, checklists, evaluation-first.https://binh-ai.com/en/posts/context-engineering-for-agentshttps://binh-ai.com/en/posts/context-engineering-for-agentsThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsAnthropic's engineering post summarized: context rot, right-altitude system prompts, just-in-time retrieval, compaction, notes and sub-agents.https://binh-ai.com/en/posts/writing-tools-for-agentshttps://binh-ai.com/en/posts/writing-tools-for-agentsThu, 29 Oct 2026 00:00:00 +0000AI Agents & SkillsAnthropic's process and five principles for effective tools/MCP: consolidate by task, namespace, meaningful results, token efficiency, clear specs.https://binh-ai.com/en/posts/advanced-tool-use-tool-search-ptchttps://binh-ai.com/en/posts/advanced-tool-use-tool-search-ptcThu, 29 Oct 2026 00:00:00 +0000Claude APIThree Claude API beta features: load tools on demand, let code call tools and filter data, teach usage by example. How to enable and when.https://binh-ai.com/en/posts/context-editing-clear-tool-resultshttps://binh-ai.com/en/posts/context-editing-clear-tool-resultsThu, 29 Oct 2026 00:00:00 +0000Claude APIclear_tool_uses and clear_thinking strategies, trigger/keep/clear_at_least parameters, prompt caching interaction and the memory tool.https://binh-ai.com/en/posts/karpathy-claude-md-four-principleshttps://binh-ai.com/en/posts/karpathy-claude-md-four-principlesThu, 29 Oct 2026 00:00:00 +0000Claude CodeThe andrej-karpathy-skills repo explained: four principles that make Claude Code guess less, write less, and produce cleaner diffs. Setup and tradeoffs.https://binh-ai.com/en/posts/claude-skills-20-best-reposhttps://binh-ai.com/en/posts/claude-skills-20-best-reposWed, 28 Oct 2026 00:00:00 +0000AI Agents & Skills20 top-starred GitHub repos for Claude: CLAUDE.md tips, workflow skills, token saving, UI design, writing, and skill security scanning.https://binh-ai.com/en/posts/claude-code-reduce-token-usagehttps://binh-ai.com/en/posts/claude-code-reduce-token-usageWed, 28 Oct 2026 00:00:00 +0000Claude CodeClear between tasks, pick the right model and effort, prefer CLIs over MCP, slim CLAUDE.md, filter output with hooks, delegate noisy work, be specific.https://binh-ai.com/en/posts/claude-code-long-session-usagehttps://binh-ai.com/en/posts/claude-code-long-session-usageWed, 28 Oct 2026 00:00:00 +0000Claude CodeSeven reasons usage climbs in long sessions, how to read cache stats in /usage, the f fix shortcut, and telling the different limit messages apart.https://binh-ai.com/en/posts/claude-code-give-claude-a-checkhttps://binh-ai.com/en/posts/claude-code-give-claude-a-checkWed, 28 Oct 2026 00:00:00 +0000Claude CodeWithout something to run, you are the test suite. Four levels of verification from prompt to Stop hook, and when a CLAUDE.md line should become a hook.https://binh-ai.com/en/posts/claude-md-keep-shorthttps://binh-ai.com/en/posts/claude-md-keep-shortWed, 28 Oct 2026 00:00:00 +0000Claude CodeCLAUDE.md loads every session, so length hurts. A two-question test, a lean template, and when to use rules, skills or hooks instead.https://binh-ai.com/en/posts/claude-prompting-core-principleshttps://binh-ai.com/en/posts/claude-prompting-core-principlesWed, 28 Oct 2026 00:00:00 +0000TipsTreat Claude like a brilliant new hire: be explicit, explain why, give 3–5 examples, separate parts with XML tags, set a role, say what to do.https://binh-ai.com/en/posts/long-context-prompting-documentshttps://binh-ai.com/en/posts/long-context-prompting-documentsWed, 28 Oct 2026 00:00:00 +0000TipsThree official techniques for long inputs: documents above the query, XML-wrapped documents with sources, and quote extraction before the task.https://binh-ai.com/en/posts/prompt-caching-api-guidehttps://binh-ai.com/en/posts/prompt-caching-api-guideWed, 28 Oct 2026 00:00:00 +0000Claude APIAutomatic vs explicit breakpoints, write/read pricing, 5-minute and 1-hour TTLs, what invalidates the cache and why caches never hit.https://binh-ai.com/en/posts/claude-code-prompt-what-it-cant-seehttps://binh-ai.com/en/posts/claude-code-prompt-what-it-cant-seeWed, 28 Oct 2026 00:00:00 +0000Claude CodePrompt tips from the best practices: symptoms and definition of done, @file and pipes, “be ambitious”, and an interview prompt that yields SPEC.md.https://binh-ai.com/en/posts/claude-code-custom-slash-commandshttps://binh-ai.com/en/posts/claude-code-custom-slash-commandsWed, 28 Oct 2026 00:00:00 +0000Claude CodeCommands in .claude/commands, frontmatter, $ARGUMENTS/$0 arguments, shell injection with !`cmd`, and narrow allowed-tools.https://binh-ai.com/en/posts/claude-code-plan-modehttps://binh-ai.com/en/posts/claude-code-plan-modeWed, 28 Oct 2026 00:00:00 +0000Claude CodeTurning on plan mode (Shift+Tab, /plan, defaultMode), the approval flow, editing plans with Ctrl+G and when to use it.https://binh-ai.com/en/posts/claude-adaptive-thinking-efforthttps://binh-ai.com/en/posts/claude-adaptive-thinking-effortWed, 28 Oct 2026 00:00:00 +0000Claude APIManual extended thinking is rejected from Claude 4.7 on. How to migrate to adaptive thinking, pick an effort level and avoid cost traps.https://binh-ai.com/en/posts/claude-api-errors-rate-limitshttps://binh-ai.com/en/posts/claude-api-errors-rate-limitsWed, 28 Oct 2026 00:00:00 +0000Claude APIWhich errors to retry, how RPM/ITPM/OTPM work, why cache reads don't count toward ITPM, and the spend-cap 429 trap.https://binh-ai.com/en/posts/serena-mcp-semantic-code-toolshttps://binh-ai.com/en/posts/serena-mcp-semantic-code-toolsWed, 28 Oct 2026 00:00:00 +0000ToolsSerena uses language servers so agents find definitions and references and edit exact symbols, cutting tokens on large repos.https://binh-ai.com/en/posts/browser-use-ai-browser-agenthttps://binh-ai.com/en/posts/browser-use-ai-browser-agentWed, 28 Oct 2026 00:00:00 +0000ToolsOpen-source Python library that lets an LLM click, type and read the web. Setup, a Claude example, use cases and security risks.https://binh-ai.com/en/posts/litellm-ai-gatewayhttps://binh-ai.com/en/posts/litellm-ai-gatewayWed, 28 Oct 2026 00:00:00 +0000ToolsCall 100+ LLMs through one format, issue budgeted, rate-limited virtual keys per team, with fallbacks and central logging.https://binh-ai.com/en/posts/promptfoo-llm-eval-cihttps://binh-ai.com/en/posts/promptfoo-llm-eval-ciWed, 28 Oct 2026 00:00:00 +0000AI Automationpromptfooconfig.yaml, assertion types, multi-model comparison and running prompt evals in CI like unit tests.https://binh-ai.com/en/posts/cheap-claude-api-proxy-riskshttps://binh-ai.com/en/posts/cheap-claude-api-proxy-risksMon, 26 Oct 2026 00:00:00 +0000Claude APIGroups often advertise Claude keys/endpoints at 50–80% off. Before plugging one into a project, understand the data, model-quality and terms risks.https://binh-ai.com/en/posts/claude-message-batches-apihttps://binh-ai.com/en/posts/claude-message-batches-apiMon, 26 Oct 2026 00:00:00 +0000Claude APIClassify 10,000 tickets, summarize a document archive, run evals… submit once as a batch, collect later, pay half.https://binh-ai.com/en/posts/claude-subscription-vs-apihttps://binh-ai.com/en/posts/claude-subscription-vs-apiMon, 26 Oct 2026 00:00:00 +0000Claude APISame models, two completely different billing models. Pick wrong and you either overpay or hit limits right when you need them.https://binh-ai.com/en/posts/subagent-model-routing-save-tokenshttps://binh-ai.com/en/posts/subagent-model-routing-save-tokensMon, 26 Oct 2026 00:00:00 +0000Claude CodeEvery Claude Code subagent can run its own model. Configure it well and you save tokens without losing quality.https://binh-ai.com/en/posts/claude-code-agent-teamshttps://binh-ai.com/en/posts/claude-code-agent-teamsMon, 26 Oct 2026 00:00:00 +0000Claude CodeOne lead coordinates, teammates have their own context, share a task list and message each other directly. Experimental, powerful, token-hungry.https://binh-ai.com/en/posts/api-key-security-ai-appshttps://binh-ai.com/en/posts/api-key-security-ai-appsMon, 26 Oct 2026 00:00:00 +0000Claude APIKeys leaked on GitHub or in client-side JavaScript get scraped and abused within minutes. A short checklist to avoid being the victim.https://binh-ai.com/en/posts/structured-output-tool-use-jsonhttps://binh-ai.com/en/posts/structured-output-tool-use-jsonMon, 26 Oct 2026 00:00:00 +0000Claude API“Only return JSON” prompts still leak prose sometimes. Define a schema as a tool and force the call — clean, parseable data.https://binh-ai.com/en/posts/mcp-skills-prompt-injection-riskshttps://binh-ai.com/en/posts/mcp-skills-prompt-injection-risksMon, 26 Oct 2026 00:00:00 +0000TipsAgents read web pages, emails, GitHub issues… and any of that text can carry hidden instructions. Simple defensive rules for Claude Code users.https://binh-ai.com/en/posts/claude-code-hooks-basicshttps://binh-ai.com/en/posts/claude-code-hooks-basicsMon, 26 Oct 2026 00:00:00 +0000Claude CodeCLAUDE.md is advice; hooks are law. Scripts run at fixed points in the agent loop, so the result is guaranteed.https://binh-ai.com/en/posts/claude-code-headless-automationhttps://binh-ai.com/en/posts/claude-code-headless-automationMon, 26 Oct 2026 00:00:00 +0000Claude CodeNot just a terminal chat — headless mode turns Claude Code into a CLI command you can drop into automated pipelines.https://binh-ai.com/en/posts/openclaw-self-hosted-ai-agenthttps://binh-ai.com/en/posts/openclaw-self-hosted-ai-agentMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsOne of the fastest-growing open-source personal-agent projects (formerly Clawdbot, Moltbot). Powerful, but it needs a safe setup.https://binh-ai.com/en/posts/openspec-spec-driven-developmenthttps://binh-ai.com/en/posts/openspec-spec-driven-developmentMon, 26 Oct 2026 00:00:00 +0000ToolsEach change is a folder with proposal, specs, design and tasks. The agent follows the tasks, then you archive and update the main spec.https://binh-ai.com/en/posts/design-md-for-ai-agentshttps://binh-ai.com/en/posts/design-md-for-ai-agentsMon, 26 Oct 2026 00:00:00 +0000Toolsawesome-design-md and getdesign.md collect 550+ DESIGN.md files analyzed from well-known sites — colors, fonts, spacing, components.https://binh-ai.com/en/posts/blender-mcp-ai-3dhttps://binh-ai.com/en/posts/blender-mcp-ai-3dMon, 26 Oct 2026 00:00:00 +0000ToolsAn addon opens a socket inside Blender; an MCP server relays Claude's commands: create objects, edit materials, pull Poly Haven assets, run Python.https://binh-ai.com/en/posts/stanford-cs146s-modern-software-developerhttps://binh-ai.com/en/posts/stanford-cs146s-modern-software-developerMon, 26 Oct 2026 00:00:00 +0000ReadingA course on how AI is changing the whole software lifecycle — IDEs, terminals, review, testing — and when AI can be trusted.https://binh-ai.com/en/posts/github-mcp-server-toolsetshttps://binh-ai.com/en/posts/github-mcp-server-toolsetsMon, 26 Oct 2026 00:00:00 +0000ToolsGitHub's own MCP server, available remote or via Docker. Toolsets and read-only mode keep context small and access safe.https://binh-ai.com/en/posts/playwright-mcp-browser-automationhttps://binh-ai.com/en/posts/playwright-mcp-browser-automationMon, 26 Oct 2026 00:00:00 +0000ToolsMicrosoft's MCP server lets Claude open pages, click, fill forms and test UIs — using page structure, no vision model required.https://binh-ai.com/en/posts/context7-up-to-date-docshttps://binh-ai.com/en/posts/context7-up-to-date-docsMon, 26 Oct 2026 00:00:00 +0000ToolsAdd “use context7” to a prompt and the agent pulls version-specific docs for your libraries instead of guessing from training data.https://binh-ai.com/en/posts/claude-code-github-actionhttps://binh-ai.com/en/posts/claude-code-github-actionMon, 26 Oct 2026 00:00:00 +0000Claude CodeAnthropic's official action runs on your own runner: automated PR review, issue triage, docs sync, scheduled maintenance.https://binh-ai.com/en/posts/claude-agent-sdk-pythonhttps://binh-ai.com/en/posts/claude-agent-sdk-pythonMon, 26 Oct 2026 00:00:00 +0000Claude APIThe same agent loop, tools, permissions and hooks as Claude Code — called from your code, with custom tools written in Python.https://binh-ai.com/en/posts/n8n-self-hosting-basicshttps://binh-ai.com/en/posts/n8n-self-hosting-basicsMon, 26 Oct 2026 00:00:00 +0000AI AutomationRun n8n with Docker on a laptop or VPS, persist data, add HTTPS — and what to do before running it for real.https://binh-ai.com/en/posts/n8n-ai-agent-node-claudehttps://binh-ai.com/en/posts/n8n-ai-agent-node-claudeMon, 26 Oct 2026 00:00:00 +0000AI AutomationCombine the Anthropic Chat Model, conversation memory and tools (HTTP, Google Sheets, databases) into an agent that runs inside a workflow.https://binh-ai.com/en/posts/ai-content-pipeline-human-approvalhttps://binh-ai.com/en/posts/ai-content-pipeline-human-approvalMon, 26 Oct 2026 00:00:00 +0000AI AutomationAutomate 90% of content drafting for a page or blog and keep the most important 10% for humans: deciding whether it goes out.https://binh-ai.com/en/posts/automation-first-ai-for-judgmenthttps://binh-ai.com/en/posts/automation-first-ai-for-judgmentMon, 26 Oct 2026 00:00:00 +0000AI AutomationA common mistake: putting an LLM in every step. Good workflows use code/if-else for the certain parts and AI for the fuzzy parts.https://binh-ai.com/en/posts/workflows-vs-agents-five-patternshttps://binh-ai.com/en/posts/workflows-vs-agents-five-patternsMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsPrompt chaining, routing, parallelization, orchestrator-workers, evaluator-optimizer — and when you actually need an autonomous agent.https://binh-ai.com/en/posts/internal-ai-assistant-chatbothttps://binh-ai.com/en/posts/internal-ai-assistant-chatbotMon, 26 Oct 2026 00:00:00 +0000AI Automation“How do I request leave?”, “What's the prod deploy process?” — a bot that answers from internal docs, cites sources, doesn't make things up.https://binh-ai.com/en/posts/esp32-mqtt-home-assistant-aihttps://binh-ai.com/en/posts/esp32-mqtt-home-assistant-aiMon, 26 Oct 2026 00:00:00 +0000AI AutomationA few-dollar microcontroller reads sensors, publishes over MQTT, Home Assistant automates — plus an AI assistant you talk to in plain language.https://binh-ai.com/en/posts/ai-email-ticket-triagehttps://binh-ai.com/en/posts/ai-email-ticket-triageMon, 26 Oct 2026 00:00:00 +0000AI AutomationThe most common AI automation pattern: read → classify → extract → route, with a confidence threshold so humans handle the hard cases.https://binh-ai.com/en/posts/ai-daily-digest-automationhttps://binh-ai.com/en/posts/ai-daily-digest-automationMon, 26 Oct 2026 00:00:00 +0000AI AutomationInstead of opening 10 tabs every morning, let a workflow collect sources, dedupe, rank and send a summary to Slack/Telegram/email.https://binh-ai.com/en/posts/automation-roi-what-to-automatehttps://binh-ai.com/en/posts/automation-roi-what-to-automateMon, 26 Oct 2026 00:00:00 +0000AI AutomationNot everything AI can do is worth automating. Score by frequency, time, cost of errors and data readiness.https://binh-ai.com/en/posts/agent-skills-explainedhttps://binh-ai.com/en/posts/agent-skills-explainedMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsA skill is a folder with a SKILL.md that teaches an agent one specific job. It’s an open standard used by Claude Code, Codex, Gemini CLI, Cursor and more.https://binh-ai.com/en/posts/agents-md-claude-codehttps://binh-ai.com/en/posts/agents-md-claude-codeMon, 26 Oct 2026 00:00:00 +0000Claude CodeSince 2.1.277, if a folder has no CLAUDE.md, Claude Code uses AGENTS.md — the open format already used by Codex, Cursor and others.https://binh-ai.com/en/posts/mcp-model-context-protocol-explainedhttps://binh-ai.com/en/posts/mcp-model-context-protocol-explainedMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsThe Model Context Protocol is a common standard for AI apps to reach external tools and data. Build a server once; every MCP-capable client can use it.https://binh-ai.com/en/posts/agent-harness-how-claude-code-workshttps://binh-ai.com/en/posts/agent-harness-how-claude-code-worksMon, 26 Oct 2026 00:00:00 +0000Claude CodeClaude Code, Cursor and Devin share the same “harness”: think → call a tool → observe → repeat. Understanding the loop makes you better at using agents.https://binh-ai.com/en/posts/claude-code-loop-cost-prompt-cachehttps://binh-ai.com/en/posts/claude-code-loop-cost-prompt-cacheMon, 26 Oct 2026 00:00:00 +0000Claude CodeEvery iteration resends the whole context. If iterations are spaced longer than the prompt cache lives, each one pays full price.https://binh-ai.com/en/posts/rtk-rust-token-killerhttps://binh-ai.com/en/posts/rtk-rust-token-killerMon, 26 Oct 2026 00:00:00 +0000ToolsA Rust binary that hooks into Claude Code and trims output from git, npm, pytest, docker… cutting most of the tokens from bash output.https://binh-ai.com/en/posts/caveman-terse-output-skillhttps://binh-ai.com/en/posts/caveman-terse-output-skillMon, 26 Oct 2026 00:00:00 +0000TipsA community skill that strips greetings, articles and filler from Claude’s replies while leaving code unchanged. It saves output tokens, but isn’t for every situation.https://binh-ai.com/en/posts/graphify-codebase-knowledge-graphhttps://binh-ai.com/en/posts/graphify-codebase-knowledge-graphMon, 26 Oct 2026 00:00:00 +0000ToolsCode parsed via AST, docs via Claude, saved as a queryable graph. The bigger the repo, the more tokens each question saves.https://binh-ai.com/en/posts/superpowers-skills-plan-before-codehttps://binh-ai.com/en/posts/superpowers-skills-plan-before-codeMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsInstead of “build me an app”, Superpowers walks the agent through brainstorm → plan → batched execution → TDD → verification before saying done.https://binh-ai.com/en/posts/autoharness-self-learning-skillshttps://binh-ai.com/en/posts/autoharness-self-learning-skillsMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsAn open-source plugin that captures reusable approaches from your sessions as skills — then updates, merges and prunes them over time.https://binh-ai.com/en/posts/ralph-wiggum-loophttps://binh-ai.com/en/posts/ralph-wiggum-loopMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsA “dumb but effective” technique: keep the agent running on the same prompt and stop only when a check that can’t lie says it passed.https://binh-ai.com/en/posts/hallmark-anti-ai-slop-design-skillhttps://binh-ai.com/en/posts/hallmark-anti-ai-slop-design-skillMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsPurple gradients, identical rounded cards, generic copy — Hallmark uses page structures, themes and a 65-point “slop test” to block those tells.https://binh-ai.com/en/posts/claude-howto-learning-pathhttps://binh-ai.com/en/posts/claude-howto-learning-pathMon, 26 Oct 2026 00:00:00 +0000Claude CodeInstalled Claude Code — now what? This open-source repo goes from slash commands to plugins, with diagrams, templates and self-check quizzes.https://binh-ai.com/en/posts/xtool-ios-without-machttps://binh-ai.com/en/posts/xtool-ios-without-macMon, 26 Oct 2026 00:00:00 +0000ToolsAn open-source Xcode alternative for turning SwiftPM packages into iOS apps, running on Linux, WSL, Windows and macOS.https://binh-ai.com/en/posts/tv-vibehack-debloat-android-tvhttps://binh-ai.com/en/posts/tv-vibehack-debloat-android-tvMon, 26 Oct 2026 00:00:00 +0000ToolsIt started with fixing a parent’s TV: a prompt and a minimal launcher let AI disable ads, telemetry and preinstalled apps — plus a list of gotchas.https://binh-ai.com/en/posts/hyperframes-video-as-codehttps://binh-ai.com/en/posts/hyperframes-video-as-codeMon, 26 Oct 2026 00:00:00 +0000ToolsHeyGen’s open-source framework: compose with HTML/CSS, render with headless Chrome + FFmpeg, plus 21 skills so agents can produce demo and launch videos.https://binh-ai.com/en/posts/remotion-agent-skillshttps://binh-ai.com/en/posts/remotion-agent-skillsMon, 26 Oct 2026 00:00:00 +0000ToolsRemotion’s official skill set guides agents through compositions, captions, maps and rendering using best practices instead of guessing the API.https://binh-ai.com/en/posts/motion-graphics-repos-for-claudehttps://binh-ai.com/en/posts/motion-graphics-repos-for-claudeMon, 26 Oct 2026 00:00:00 +0000Reading“20 motion graphics repos” lists are easy to misread: some are skills, some are video engines, some are just web animation libraries.https://binh-ai.com/en/posts/rea-reverse-engineer-anything-reality-checkhttps://binh-ai.com/en/posts/rea-reverse-engineer-anything-reality-checkMon, 26 Oct 2026 00:00:00 +0000AI Agents & SkillsREA helps AI agents investigate how binaries, APKs and web apps behave. It’s real and useful — but the clickbait overstates it and skips the legal and safety risks.https://binh-ai.com/en/posts/windows-event-4625-brute-force-triagehttps://binh-ai.com/en/posts/windows-event-4625-brute-force-triageMon, 26 Oct 2026 00:00:00 +0000System DesignRed logs don’t always mean an attack. How SOC analysts triage Windows logon failures: which fields matter, when it’s brute force, when it’s a stale password.https://binh-ai.com/en/posts/cloudflare-security-audit-skillhttps://binh-ai.com/en/posts/cloudflare-security-audit-skillMon, 26 Oct 2026 00:00:00 +0000ToolsVibe coding is fast but leaky. Cloudflare’s new skill scans your codebase and verifies whether each vulnerability is real before you ship.https://binh-ai.com/en/posts/bloom-filter-false-positivehttps://binh-ai.com/en/posts/bloom-filter-false-positiveMon, 26 Oct 2026 00:00:00 +0000System DesignA tiny bit array in RAM lets RocksDB and Cassandra skip countless pointless disk reads — at the cost of the occasional false alarm.https://binh-ai.com/en/posts/replication-lag-read-your-writeshttps://binh-ai.com/en/posts/replication-lag-read-your-writesMon, 26 Oct 2026 00:00:00 +0000System DesignNothing was lost — you’re just reading from a replica that hasn’t caught up yet. The cleanest fix: read-your-writes.https://binh-ai.com/en/posts/claude-code-desktop-multitaskhttps://binh-ai.com/en/posts/claude-code-desktop-multitaskMon, 26 Oct 2026 00:00:00 +0000Claude CodeA session sidebar, the /btw command, automatic sub-tasks and saved Environments — four changes that make multi-threaded work easier.https://binh-ai.com/en/posts/tcp-handshake-first-request-rtthttps://binh-ai.com/en/posts/tcp-handshake-first-request-rttMon, 26 Oct 2026 00:00:00 +0000System DesignFirst request 160ms, next ones just 80ms — even with an instant server. The culprit is the SYN / SYN-ACK / ACK greeting.https://binh-ai.com/en/posts/captist-auto-zoom-screen-recorderhttps://binh-ai.com/en/posts/captist-auto-zoom-screen-recorderMon, 26 Oct 2026 00:00:00 +0000ToolsShip demo videos, tutorials and progress updates faster: auto-zoom, sticky redaction, keystroke overlays and local processing.https://binh-ai.com/en/posts/debounce-search-inputhttps://binh-ai.com/en/posts/debounce-search-inputMon, 26 Oct 2026 00:00:00 +0000TipsA search box that calls the API on every keystroke floods the server with half-typed queries. Debounce waits until you’re done.https://binh-ai.com/en/posts/ambient-light-for-youtubehttps://binh-ai.com/en/posts/ambient-light-for-youtubeMon, 26 Oct 2026 00:00:00 +0000ToolsAn open-source extension that uses WebGL to cast a glow around YouTube videos — no LED strips required.https://binh-ai.com/en/posts/animated-avatars-ai-chatbot-uxhttps://binh-ai.com/en/posts/animated-avatars-ai-chatbot-uxMon, 26 Oct 2026 00:00:00 +0000AI AutomationSwap the boring loading spinner for a character that changes motion with the agent’s Idle, Working and Sleeping states.https://binh-ai.com/en/posts/liquid-glass-webgl-uihttps://binh-ai.com/en/posts/liquid-glass-webgl-uiMon, 26 Oct 2026 00:00:00 +0000ToolsCSS backdrop-filter only gives a flat blur. Liquid Glass libraries use WebGL to compute refraction for a convincing glass look.https://binh-ai.com/en/posts/iphone-duo-mockup-generatorhttps://binh-ai.com/en/posts/iphone-duo-mockup-generatorMon, 26 Oct 2026 00:00:00 +0000ToolsDays after launch, the community already has dual-screen mockup tools: upload your UI, export high-quality images.https://binh-ai.com/en/posts/page-mascot-cursor-trackinghttps://binh-ai.com/en/posts/page-mascot-cursor-trackingMon, 26 Oct 2026 00:00:00 +0000ToolsAn interactive mascot library that makes static pages feel alive: characters track the cursor, react to clicks, and AI agents can draw custom ones.https://binh-ai.com/en/posts/morphicons-svg-icon-morphinghttps://binh-ai.com/en/posts/morphicons-svg-icon-morphingMon, 26 Oct 2026 00:00:00 +0000ToolsTurn a Menu button into an X with a shape morph instead of a rotate or fade. Works with Lucide, Tabler and Heroicons, with zero heavy dependencies.https://binh-ai.com/en/posts/retry-storm-exponential-backoff-jitterhttps://binh-ai.com/en/posts/retry-storm-exponential-backoff-jitterMon, 26 Oct 2026 00:00:00 +0000System Design12 clients retrying at once knock over a 6-slot server. Backoff decides how long to wait; jitter decides whether the server survives.https://binh-ai.com/en/posts/write-ahead-log-durabilityhttps://binh-ai.com/en/posts/write-ahead-log-durabilityMon, 26 Oct 2026 00:00:00 +0000System DesignAll 6 writes were acked but only 3 reached the table — and after a power cut nothing is lost, because the log is written first and the table later.https://binh-ai.com/en/posts/circuit-breaker-patternhttps://binh-ai.com/en/posts/circuit-breaker-patternMon, 26 Oct 2026 00:00:00 +0000System DesignThe service is down, yet every call still waits 1000ms for an error. A circuit breaker counts failures, trips, and fails fast.https://binh-ai.com/en/posts/event-loop-settimeout-zerohttps://binh-ai.com/en/posts/event-loop-settimeout-zeroMon, 26 Oct 2026 00:00:00 +0000Tips“0ms” doesn’t promise to run now — only to queue early. Microtasks always go before macrotasks.https://binh-ai.com/en/posts/backpressure-queue-watermarkshttps://binh-ai.com/en/posts/backpressure-queue-watermarksMon, 26 Oct 2026 00:00:00 +0000System DesignThe producer pushes 3 msg/s, the consumer handles 1. Without backpressure the queue grows until it runs out of memory. With it, the consumer sets the pace.https://binh-ai.com/en/posts/p2p-cdn-webrtc-video-streaminghttps://binh-ai.com/en/posts/p2p-cdn-webrtc-video-streamingMon, 26 Oct 2026 00:00:00 +0000System DesignThe CDN sends only a few original copies; viewers’ players pass the rest to each other — and fall back to the CDN the moment a peer leaves.https://binh-ai.com/en/posts/database-connection-poolhttps://binh-ai.com/en/posts/database-connection-poolMon, 26 Oct 2026 00:00:00 +0000System DesignThe query takes 12ms but the API is still slow? Connection setup is often the culprit. A pool removes that cost — until load exceeds it and requests queue.https://binh-ai.com/en/posts/multi-layer-caching-stackhttps://binh-ai.com/en/posts/multi-layer-caching-stackMon, 26 Oct 2026 00:00:00 +0000System DesignBrowser, CDN, Redis, database — stacked. A request pays only down to the layer that answers, and every miss teaches the stack.