https://www.frr.dev/Recent content on frr.devHugo -- 0.161.1en-USMon, 01 Jun 2026 14:54:58 +0200https://www.frr.dev/posts/cli-design-ai-agents-errors/Fri, 22 May 2026 09:00:00 +0200https://www.frr.dev/posts/cli-design-ai-agents-errors/An AI agent made over 500 errors using a CLI across 165 sessions. Its error log serves as a reverse blueprint for the interface you should have designed.https://www.frr.dev/posts/cloudflare-ask-ai-api-token/Thu, 21 May 2026 11:30:00 +0200https://www.frr.dev/posts/cloudflare-ask-ai-api-token/Cloudflare's in-dashboard AI assistant silently provisioned a permanent, read-everything API token across all my accounts, zones and users.https://www.frr.dev/posts/sentry-frontend-brave-shields-tunnel/Mon, 18 May 2026 18:30:00 +0200https://www.frr.dev/posts/sentry-frontend-brave-shields-tunnel/Brave Shields blocks Sentry even with shields down. Learn how to fix it with your own Flask tunnel and avoid three undocumented gotchas.https://www.frr.dev/posts/agentic-experience-data-driven-cli-design-llm/Wed, 29 Apr 2026 19:00:00 +0100https://www.frr.dev/posts/agentic-experience-data-driven-cli-design-llm/I analyzed 1,324 real calls from Claude Code to my Linear CLI. 15.9% failed. The errors reveal how to design tools that agents can use without breaking.https://www.frr.dev/posts/two-tier-pipeline-deterministic-coreml/Sat, 18 Apr 2026 20:00:00 +0200https://www.frr.dev/posts/two-tier-pipeline-deterministic-coreml/The same architectural pattern solves sentiment analysis and image cleanup: deterministic first, CoreML only for what remains. PyTorch → ONNX → Apple Neural Engine in three commands.https://www.frr.dev/posts/perceptual-hash-adversarial-collisions/Sat, 18 Apr 2026 18:00:00 +0200https://www.frr.dev/posts/perceptual-hash-adversarial-collisions/In August 2021, researchers broke NeuralHash in less than two weeks. The attack wasn’t a fluke—it’s a structural property of the perceptual hash family. The EU is legislating on this.https://www.frr.dev/posts/perceptual-hash-eu-chat-control/Sat, 18 Apr 2026 12:00:00 +0200https://www.frr.dev/posts/perceptual-hash-eu-chat-control/The EU Chat Control and Apple’s failed on-device scanning rely on the same 40-line Python algorithm. Understanding how it works is essential for informed discussion.https://www.frr.dev/posts/cli-agent-experience-design-llm-first/Tue, 07 Apr 2026 20:00:00 +0200https://www.frr.dev/posts/cli-agent-experience-design-llm-first/Agent Experience (AX) principles are changing how CLIs are designed when the primary user is an LLM. Output that’s token-efficient, fuzzy matching, actionable errors, and zero mandatory flags.https://www.frr.dev/posts/apple-on-device-model-terrible-chat-great-structured-output/Tue, 07 Apr 2026 11:00:00 +0200https://www.frr.dev/posts/apple-on-device-model-terrible-chat-great-structured-output/I built a macOS app to stress-test every capability of Apple Intelligence on-device. The 3B model is mediocre at free-form text, but with constrained decoding and tool calling, the results are surprisingly good.https://www.frr.dev/posts/apple-nltagger-sentiment-bias-developer-text/Sun, 05 Apr 2026 11:00:00 +0200https://www.frr.dev/posts/apple-nltagger-sentiment-bias-developer-text/Apple's NLTagger scores 'delete the temp file' as -0.8 (very negative). It's systematic bias in technical text that nobody has documented. Data, causes, and an open source solution.https://www.frr.dev/posts/libfoundationmodels-apple-llm-c-ffi-bridge/Sun, 05 Apr 2026 11:00:00 +0200https://www.frr.dev/posts/libfoundationmodels-apple-llm-c-ffi-bridge/libfoundationmodels exposes Apple's on-device LLM as a 4-function C dylib. The async Swift → sync C bridge using Mutex and DispatchSemaphore, and why this is more useful than a local HTTP server.https://www.frr.dev/posts/foundation-hooks-on-device-llm-git-issue-tracking/Sun, 05 Apr 2026 11:00:00 +0200https://www.frr.dev/posts/foundation-hooks-on-device-llm-git-issue-tracking/Coding agents use $15/M token cloud models for commit messages. foundation-hooks replaces them with Apple's on-device LLM: 300ms, 0 tokens, 0 cost. Open source repo with make install.https://www.frr.dev/posts/turboquant-one-month-later-implementations-controversy-benchmarks/Sun, 05 Apr 2026 11:00:00 +0200https://www.frr.dev/posts/turboquant-one-month-later-implementations-controversy-benchmarks/Google's TurboQuant paper got 575 points on HN and crashed memory stocks. One month later: which implementations actually work, the RaBitQ controversy Google won't address, and how to run it on your Mac today.https://www.frr.dev/posts/apple-intelligence-free-llm-dev-tooling/Sat, 04 Apr 2026 11:00:00 +0200https://www.frr.dev/posts/apple-intelligence-free-llm-dev-tooling/Apple Intelligence includes a 3B-parameter model on your Mac. Learn how to use it for classification, naming, triage, and commit messages without paying for APIs. Copy-pasteable Swift recipes.https://www.frr.dev/posts/ai-coding-router-experiment/Mon, 30 Mar 2026 20:31:34 +0200https://www.frr.dev/posts/ai-coding-router-experiment/My plan to distribute work among Claude, Codex, and GLM/Z.AI without automating too early, cutting costs without losing judgment.https://www.frr.dev/posts/codex-skills-not-slash-commands/Mon, 30 Mar 2026 18:30:00 +0200https://www.frr.dev/posts/codex-skills-not-slash-commands/If you're coming from Claude Code, this will save you some serious frustration: in Codex, a skill is not a slash command. They're used for different purposes.https://www.frr.dev/posts/nltagger-sentiment-bias-technical-text/Sat, 28 Mar 2026 12:00:00 +0100https://www.frr.dev/posts/nltagger-sentiment-bias-technical-text/Apple's NLTagger scores perfectly neutral technical text as negative. An empirical analysis of bias, why it happens, and what to use instead.https://www.frr.dev/posts/claude-mornings-codex-afternoons-two-agent-workflow/Thu, 26 Mar 2026 22:00:00 +0100https://www.frr.dev/posts/claude-mornings-codex-afternoons-two-agent-workflow/165 sessions with Claude Code and 27 with Codex CLI reveal a clear pattern: each agent excels in different tasks. Learn how to organize your day with two AI copilots.https://www.frr.dev/posts/python-tooling-rust-uv-ruff-ferris-was-right/Thu, 26 Mar 2026 21:00:00 +0100https://www.frr.dev/posts/python-tooling-rust-uv-ruff-ferris-was-right/Python has had a fragmented and slow tooling ecosystem for years. The revolution came from where nobody expected: Rust. uv, Ruff and ty have changed the game. The Cult of Ferris was right.https://www.frr.dev/posts/skill-before-after-tolerant-tool-fewer-instructions/Thu, 26 Mar 2026 18:00:00 +0100https://www.frr.dev/posts/skill-before-after-tolerant-tool-fewer-instructions/My Linear skill had 150 lines of workarounds: hardcoded UUIDs, curl fallbacks, notes like 'the CLI doesn’t support X.' I didn’t rewrite them — I made them unnecessary by building a tolerant tool.https://www.frr.dev/posts/madness-driven-design-don-quixote-sancho-panza-ai-copilot/Thu, 26 Mar 2026 18:00:00 +0100https://www.frr.dev/posts/madness-driven-design-don-quixote-sancho-panza-ai-copilot/MDD: a methodology for designing tools that absorb the hallucinations of your AI copilot. Two layers—error archaeology and adversarial verification—inspired by Don Quixote and Sancho Panza.https://www.frr.dev/posts/cli-output-xml-toon-token-efficient-llm/Thu, 26 Mar 2026 18:00:00 +0100https://www.frr.dev/posts/cli-output-xml-toon-token-efficient-llm/XML repeats tags, JSON repeats keys. When your consumer is an LLM, every token counts. The story of how I ended up with a compact positional format and discovered it already existed: TOON.https://www.frr.dev/posts/adversarial-programming-ai-copilot-invents-api/Thu, 26 Mar 2026 11:00:00 +0100https://www.frr.dev/posts/adversarial-programming-ai-copilot-invents-api/Your AI generates plausible code against non-existent APIs. Schema introspection, real fixtures, and fetch/process separation: the anti-hallucination methodology that works.https://www.frr.dev/posts/linear-agent-cli-rust-agent-already-had/Wed, 25 Mar 2026 19:00:00 +0100https://www.frr.dev/posts/linear-agent-cli-rust-agent-already-had/Linear launched their AI agent. Sounds great, but what we really needed was a functional CLI that didn’t suck. And if we have to rewrite it, let’s do it in Rust.https://www.frr.dev/posts/transform-conquer-turboquant-polar-coordinates-llm-compression/Wed, 25 Mar 2026 11:00:00 +0100https://www.frr.dev/posts/transform-conquer-turboquant-polar-coordinates-llm-compression/TurboQuant compresses LLM KV cache 6x without losing precision. The trick: don't invent a better compressor, move data to polar coordinates where compression becomes trivial.https://www.frr.dev/posts/type-system-linter-llm-stringly-typed-code/Mon, 23 Mar 2026 21:00:00 +0100https://www.frr.dev/posts/type-system-linter-llm-stringly-typed-code/LLMs generate code that compiles but abuses strings, optionals, and nil where specific types should exist. How to detect it with a linter and an auditing skill.https://www.frr.dev/posts/plan-md-second-opinion-claude-codex-review/Mon, 23 Mar 2026 19:00:00 +0100https://www.frr.dev/posts/plan-md-second-opinion-claude-codex-review/A developer got Claude and Codex to argue about their plan until it was solid. The idea is brilliant: use a second model as an adversarial reviewer. 14 bugs caught, zero human intervention.https://www.frr.dev/posts/mole-mo-purge-analyze-mac-cleanup/Mon, 23 Mar 2026 13:00:00 +0100https://www.frr.dev/posts/mole-mo-purge-analyze-mac-cleanup/Mole promises to clean and optimize your Mac. After testing it all, only two commands are worth your time: mo purge and mo analyze. The rest is just fluff.https://www.frr.dev/posts/zoxide-ai-agent-memory-project-paths/Sat, 14 Mar 2026 16:00:00 +0100https://www.frr.dev/posts/zoxide-ai-agent-memory-project-paths/Your AI agent doesn't remember where your projects are. zoxide and CLAUDE.md are the solution: a smart cd and a memory tattoo so they stop getting lost.https://www.frr.dev/posts/bayesian-estimator-adaptive-polling-battery/Thu, 12 Mar 2026 19:00:00 +0100https://www.frr.dev/posts/bayesian-estimator-adaptive-polling-battery/How a scalar Kalman filter reduces network calls by 80% in a menu bar app, adapts to battery status, and sleeps on its own when you're inactive.https://www.frr.dev/posts/claude-code-ghostty-worktrees-mac-setup/Wed, 11 Mar 2026 23:55:00 +0100https://www.frr.dev/posts/claude-code-ghostty-worktrees-mac-setup/How I combine Ghostty, git worktrees, and Claude Code in my daily macOS development workflow. No flickers, massive scrollback, and true parallelism with worktrees.https://www.frr.dev/posts/codex-cli-worktrees-manual-parallelism/Wed, 11 Mar 2026 23:45:00 +0100https://www.frr.dev/posts/codex-cli-worktrees-manual-parallelism/Codex CLI doesn’t support --worktree or --tmux. But with manual Git worktrees and a few tricks, you can achieve real parallelism. A practical guide with the bugs you’ll encounter.https://www.frr.dev/posts/claude-code-worktrees-parallel-agents-tips/Wed, 11 Mar 2026 23:30:00 +0100https://www.frr.dev/posts/claude-code-worktrees-parallel-agents-tips/Claude Code has native support for git worktrees: --worktree, isolated sub-agents, custom agents with isolation, --tmux for parallel sessions. Five tips from Boris Cherny.https://www.frr.dev/posts/boring-discipline-beats-magic-roundup/Wed, 11 Mar 2026 23:00:00 +0100https://www.frr.dev/posts/boring-discipline-beats-magic-roundup/PostgreSQL without sharding, a while loop without architecture, cron without cloud platforms. Six articles saying the same thing: boring discipline works better than flashy technology.https://www.frr.dev/posts/codex-cli-autonomous-agent-two-flags/Wed, 11 Mar 2026 22:00:00 +0100https://www.frr.dev/posts/codex-cli-autonomous-agent-two-flags/Tired of Codex CLI asking permission for every file? Use --full-auto or -a never -s workspace-write for autonomous agent mode. Step-by-step guide with Claude Code comparison.https://www.frr.dev/posts/context-engineering-invisible-skill-ai-agents/Wed, 11 Mar 2026 22:00:00 +0100https://www.frr.dev/posts/context-engineering-invisible-skill-ai-agents/Prompt engineering is writing a good prompt. Context engineering is designing EVERYTHING the model sees: what goes in, in what order, what gets discarded, what gets compressed. And it's what really matters.https://www.frr.dev/posts/email-bankruptcy-memento-reminders/Wed, 11 Mar 2026 21:30:00 +0100https://www.frr.dev/posts/email-bankruptcy-memento-reminders/Email is an inbox anyone can fill with junk. I've declared bankruptcy multiple times. This time, instead of promising to manage my inbox better, I’ve decided to stop using it as a reminder system.https://www.frr.dev/posts/launch-council-mvp-llm-adversarial-review/Wed, 11 Mar 2026 21:00:00 +0100https://www.frr.dev/posts/launch-council-mvp-llm-adversarial-review/I designed a council of five simulated LLM experts to evaluate MVPs before writing code. Paul Graham, Lessig, Godin, Balaji, DHH — and why each one is part of the team.https://www.frr.dev/posts/diy-codex-automations-claude-code-systemd/Wed, 11 Mar 2026 20:00:00 +0100https://www.frr.dev/posts/diy-codex-automations-claude-code-systemd/A practical tutorial to replicate OpenAI's Codex Automations using Claude Code, systemd timers, and Gitea. Agents that work while you sleep, without relying on any desktop app.https://www.frr.dev/posts/openai-postgresql-800m-users-single-writer/Wed, 11 Mar 2026 20:00:00 +0100https://www.frr.dev/posts/openai-postgresql-800m-users-single-writer/OpenAI serves 800M ChatGPT users with a single PostgreSQL primary and ~50 replicas. No sharding, no microservices. Deliberate simplicity beats over-engineering.https://www.frr.dev/posts/ai-coding-agent-anatomy-while-loop/Wed, 11 Mar 2026 20:00:00 +0100https://www.frr.dev/posts/ai-coding-agent-anatomy-while-loop/Codex CLI and Claude Code look like magic, but inside they're just a loop with 5 phases, an LLM, and tools. We break down the agentic loop step by step with diagrams.https://www.frr.dev/posts/reverse-engineer-neural-network-senior-debugging/Wed, 11 Mar 2026 19:00:00 +0100https://www.frr.dev/posts/reverse-engineer-neural-network-senior-debugging/Jane Street hid MD5 inside a neural network with integer weights. The process to uncover it is a masterclass in debugging that every senior should study.https://www.frr.dev/posts/llm-cache-cost-counterintuitive-savings/Tue, 10 Mar 2026 18:00:00 +0100https://www.frr.dev/posts/llm-cache-cost-counterintuitive-savings/Prompt caching reduces input cost by 90%, but cache writes cost 25% extra. Your bill goes up before it goes down. Here's how it works and how to manage your expenses.https://www.frr.dev/posts/simplify-jedi-council-ai-code-review/Mon, 09 Mar 2026 18:00:00 +0100https://www.frr.dev/posts/simplify-jedi-council-ai-code-review//simplify by Claude Code reviews your code with three generic agents. It's okay, but it creates noise. I replaced it with a Jedi Council: Kent Beck, Martin Fowler, and Mike Acton with real decision rules. Fewer findings, zero false positives, and one bug the other tool missed.https://www.frr.dev/posts/claude-code-loop-vs-cron-scheduling/Mon, 09 Mar 2026 12:00:00 +0100https://www.frr.dev/posts/claude-code-loop-vs-cron-scheduling/Claude Code 2.1.71 introduces /loop, an inline scheduler for recurring tasks. What it does, how to use it, and why it doesn’t quite replace a proper cron job.https://www.frr.dev/posts/ztrace-xctrace-compact-summary-llm/Sun, 08 Mar 2026 14:00:00 +0100https://www.frr.dev/posts/ztrace-xctrace-compact-summary-llm/xctrace exports 33,000-line XML files that blow up any LLM's context window. ztrace condenses them into 10 actionable lines. How and why.https://www.frr.dev/posts/impossible-path-ai-agent-guardrails/Fri, 27 Feb 2026 20:00:00 +0100https://www.frr.dev/posts/impossible-path-ai-agent-guardrails/When an AI agent operates your ETL pipeline, forbidding things doesn't work. Only making the wrong path structurally impossible works.https://www.frr.dev/posts/claude-code-remote-control-hidden-api/Fri, 27 Feb 2026 18:00:00 +0100https://www.frr.dev/posts/claude-code-remote-control-hidden-api/Anthropic announced Remote Control for Claude Code. We investigated the bridge API, WebSockets, on-disk traces, and why it still doesn't work for all Max accounts in February 2026.https://www.frr.dev/posts/rustyclaw-manifesto-rewrite-ai-agent-rust/Tue, 24 Feb 2026 18:00:00 +0100https://www.frr.dev/posts/rustyclaw-manifesto-rewrite-ai-agent-rust/I’m porting 8,300 lines of Python to Rust using LLMs as copilots. The real goal: testing adversarial development in a hardcore porting process. With raw data and a hallucination counter.https://www.frr.dev/posts/llm-read-json-900-times-loop-performance/Tue, 24 Feb 2026 14:00:00 +0100https://www.frr.dev/posts/llm-read-json-900-times-loop-performance/An LLM wrote code that read and parsed a JSON file from disk on every iteration of a 900-file loop. It's a first-year computer science mistake. No linter can catch it.https://www.frr.dev/posts/git-cliff-automatic-changelog/Sun, 22 Feb 2026 22:00:00 +0100https://www.frr.dev/posts/git-cliff-automatic-changelog/How to use git-cliff to generate retroactive changelogs from conventional commits, with tags, Tera templates and a couple of tricks that aren't in the documentation.https://www.frr.dev/posts/tokamak-estimating-claude-quota-without-api/Sun, 22 Feb 2026 20:00:00 +0100https://www.frr.dev/posts/tokamak-estimating-claude-quota-without-api/Tokamak reads Claude Max quota with cookies. Anthropic just banned third-party OAuth. Here's how we prepare with statistics for the day they shut down the API.https://www.frr.dev/posts/macos-notarization-guide-linter/Sun, 22 Feb 2026 18:00:00 +0100https://www.frr.dev/posts/macos-notarization-guide-linter/What Apple notarization is, why it exists, the most common errors that will drive you crazy, and a 25-check script so you never fail again.https://www.frr.dev/posts/tart-macos-vms-from-terminal/Sat, 21 Feb 2026 14:00:00 +0100https://www.frr.dev/posts/tart-macos-vms-from-terminal/I needed to test my app on a clean macOS. UTM can't be scripted. Tart can: one brew install and a command to get a macOS VM with shared folder.https://www.frr.dev/posts/prompt-caching-99-percent-cache-hit/Thu, 19 Feb 2026 16:00:00 +0100https://www.frr.dev/posts/prompt-caching-99-percent-cache-hit/Your LLM remembers almost everything you told it 30 seconds ago. Not by magic: through KV tensors, GPU VRAM, and local SSDs. Here's how prompt caching works under the hood.https://www.frr.dev/posts/summoning-the-wise-mentoring-experts-llm/Wed, 18 Feb 2026 20:00:00 +0100https://www.frr.dev/posts/summoning-the-wise-mentoring-experts-llm/You can ask an LLM to simulate Edward Tufte critiquing your chart or Charlie Munger reviewing your budget. And it works surprisingly well.https://www.frr.dev/posts/beads-is-dead/Wed, 18 Feb 2026 14:00:00 +0100https://www.frr.dev/posts/beads-is-dead/I'm retiring Beads from my Claude Code workflow. What it brought to the table, why it stopped working, and what replaced it.https://www.frr.dev/posts/git-worktrees-coding-agents-parallel/Mon, 16 Feb 2026 18:00:00 +0100https://www.frr.dev/posts/git-worktrees-coding-agents-parallel/If you use coding agents like Claude Code, Cursor, or Copilot, a single checkout falls short. Git worktrees give you multiple working directories on the same repo without copying anything.https://www.frr.dev/posts/five-defenses-code-hallucinations/Mon, 16 Feb 2026 12:00:00 +0100https://www.frr.dev/posts/five-defenses-code-hallucinations/Detailed analysis of 5 reactive measures against LLM-generated fictitious code: contract testing, real fixtures, smoke tests, anomaly detection, and LLM-as-Judge. Which ones work and which are just theater.https://www.frr.dev/posts/silent-failure-ai-makes-stuff-up-tests-everything-fine/Fri, 13 Feb 2026 10:00:00 +0100https://www.frr.dev/posts/silent-failure-ai-makes-stuff-up-tests-everything-fine/LLMs don't just lie: they build coherent fictions with DTOs, fixtures, and tests that validate the lie. And today we have no reliable way to detect it.https://www.frr.dev/posts/memory-md-claude-code-field-notebook/Thu, 12 Feb 2026 18:00:00 +0100https://www.frr.dev/posts/memory-md-claude-code-field-notebook/Claude Code has a file where it takes notes between sessions. You don't write it. It writes it. And almost nobody knows it exists.https://www.frr.dev/posts/authorization-fatigue-1password-cache/Thu, 12 Feb 2026 15:00:00 +0100https://www.frr.dev/posts/authorization-fatigue-1password-cache/If your security tool asks for authorization every 30 seconds, you end up saying yes without looking. Here's how we created a cache for op that solves the problem without compromising security (we think).https://www.frr.dev/posts/when-your-ai-becomes-your-worst-enemy/Fri, 06 Feb 2026 19:00:00 +0100https://www.frr.dev/posts/when-your-ai-becomes-your-worst-enemy/How the inherent incentives of an LLM can turn a routine task into a disaster. Lessons from a real post-mortem.https://www.frr.dev/posts/git-secrets-1password-cli/Thu, 05 Feb 2026 12:00:00 +0100https://www.frr.dev/posts/git-secrets-1password-cli/39 million credentials leaked on GitHub in 2024. AWS, OpenAI, production tokens... all exposed. Here's how I stopped playing Russian roulette with my secrets.https://www.frr.dev/posts/apple-84gb-jellyfish-duplicated-ssd/Thu, 29 Jan 2026 12:00:00 +0100https://www.frr.dev/posts/apple-84gb-jellyfish-duplicated-ssd/I wanted to test Kimi K2 and was running out of disk space. Turns out macOS stores screensavers duplicated. Yes, two copies of 42GB of jellyfish videos.https://www.frr.dev/posts/bfs-vs-dfs-how-we-think-and-search/Wed, 28 Jan 2026 18:00:00 +0100https://www.frr.dev/posts/bfs-vs-dfs-how-we-think-and-search/BFS and DFS aren't just algorithms: they're the two ways your brain navigates the world. And your professional success depends on which one you master.https://www.frr.dev/posts/claude-code-native-build-bun/Tue, 27 Jan 2026 21:00:00 +0100https://www.frr.dev/posts/claude-code-native-build-bun/Anthropic has released a native executable of Claude Code that doesn't need Node.js. It weighs 100MB. Why? I'll explain what's inside and who really wins with this change.https://www.frr.dev/posts/half-life-skills-continuous-learning/Tue, 27 Jan 2026 12:00:00 +0100https://www.frr.dev/posts/half-life-skills-continuous-learning/Your experience has an expiration date. The data on professional obsolescence and what to do to avoid being left behind.https://www.frr.dev/posts/clawdbot-ai-assistant-space-lobster/Mon, 26 Jan 2026 19:00:00 +0100https://www.frr.dev/posts/clawdbot-ai-assistant-space-lobster/Clawdbot is the open source AI assistant that exploded on GitHub. 17,000 stars, 300 security issues, and a space lobster. Here's what it is and why it matters.https://www.frr.dev/posts/claude-vm-cowork-code-comparison/Sun, 25 Jan 2026 18:00:00 +0100https://www.frr.dev/posts/claude-vm-cowork-code-comparison/That 10.8 GB claudevm.bundle in Claude Desktop's vm_bundles folder is a hidden Ubuntu VM (Cowork). What it does, why it's there, and how to delete it safely.https://www.frr.dev/posts/terminal-gpu-battery-macbook-ghostty-iterm2/Sat, 24 Jan 2026 19:00:00 +0100https://www.frr.dev/posts/terminal-gpu-battery-macbook-ghostty-iterm2/How a 1970s-era application like a terminal emulator can consume more battery than Zoom with video. The irony, the causes, and the solutions.https://www.frr.dev/posts/tasks-vs-beads-claude-code-memory/Fri, 23 Jan 2026 12:00:00 +0100https://www.frr.dev/posts/tasks-vs-beads-claude-code-memory/Tutorial on the three tracking layers in Claude Code: Linear for strategic vision, Beads for tactical work, and Tasks for execution. When to use each one.https://www.frr.dev/posts/subagents-claude-code-tutorial/Tue, 20 Jan 2026 12:00:00 +0100https://www.frr.dev/posts/subagents-claude-code-tutorial/Tutorial on subagents in Claude Code: how to launch them, communicate with them, run them in parallel, and create your ownhttps://www.frr.dev/posts/slow-git-notebooks-fsmonitor-nbstripout/Mon, 19 Jan 2026 18:00:00 +0100https://www.frr.dev/posts/slow-git-notebooks-fsmonitor-nbstripout/Your repo with notebooks takes forever to run git status. Here's why and how to fix it with FSMonitor and nbstripout.https://www.frr.dev/posts/chromadb-tutorial-curriculum-validation/Sun, 18 Jan 2026 14:00:00 +0100https://www.frr.dev/posts/chromadb-tutorial-curriculum-validation/Practical ChromaDB tutorial: building a system that automatically verifies that educational material doesn't use concepts before teaching them. Semantic search for busy people.https://www.frr.dev/posts/bun-tutorial-devs-node/Sun, 18 Jan 2026 12:00:00 +0100https://www.frr.dev/posts/bun-tutorial-devs-node/Complete Bun tutorial for TypeScript developers already using Node, pnpm and Vitest. What it is, why Anthropic just bought it, and how to migrate without breaking everything.https://www.frr.dev/posts/claude-code-skills-tutorial/Mon, 12 Jan 2026 18:00:00 +0100https://www.frr.dev/posts/claude-code-skills-tutorial/Tutorial on creating Skills in Claude Code v2.1.3+: custom commands that your AI can execute automaticallyhttps://www.frr.dev/posts/linear-beads-ai-memory/Sun, 12 Jan 2025 17:00:00 +0100https://www.frr.dev/posts/linear-beads-ai-memory/Tutorial on using Linear for product management and Beads as persistent memory for AI agents like Claude Codehttps://www.frr.dev/about/Mon, 01 Jan 0001 00:00:00 +0000https://www.frr.dev/about/Fernando Rodríguez — CTO at KeepCoding, Guinness World Record holder, and developer writing about AI agents, Swift, Python, and building things that work.https://www.frr.dev/tokamak/privacy/Mon, 01 Jan 0001 00:00:00 +0000https://www.frr.dev/tokamak/privacy/Privacy Policy for Tokamak, the Claude Quota HUD for macOS