CI Gates, Test Result Caches, and the Code Graph That Might Tell Us What to Test
What I learned from splitting Rails CI by risk, caching exact test results, and exploring whether AI and code graphs can make test selection safer.
Category: engineering
When a Green Deployment Still Serves a 500
A REX production incident exposed the gap between a successful Kubernetes rollout and a working application, and what we changed to close it.
The One-Person Company Is a Coordination Problem
I am exploring how a one-person company can use agents, observability, and market intelligence without turning automation into an ungoverned second company.
Hermes vs OpenClaw, NanoClaw, ZeroClaw, IronClaw: Agent Architecture
A practical, architectural comparison of Hermes, OpenClaw, NanoClaw, ZeroClaw, and IronClaw, and why a layered agent system beats picking one winner.
I Audited the OpenVPN Server I Wrote a Guide For
Two years after publishing a guide to restricting OpenVPN clients to specific IPs, I audited the server it built, and most of the findings trace back to the guide.
Codebase Memory for Coding Agents: Graft, Graphify, and codebase-memory-mcp
A practical comparison of Graft, Graphify, and codebase-memory-mcp: what each indexes, how each feeds coding agents, and when to use one instead of another.
When a Rails or LMS Platform Becomes Hard to Change
A practical way for small education and SaaS teams to diagnose Rails or LMS platform friction and decide whether a rewrite is actually justified.
When Ubuntu Builds Fail on libvips but macOS Does Not
Why a Rails test suite can fail on Ubuntu because of an older native libvips, and how to upgrade it safely without confusing it with the ruby-vips gem.
Testing Small Local LLMs on Modest Hardware: A Portable Workflow
A repeatable way to download, serve, benchmark, and choose small GGUF models for Hermes and coding-agent work across different CPU and memory budgets.
Local Gemma with Hermes: Cheap Delegation for Rails Maintenance
A practical guide to using Hermes Agent with a local Gemma model for routine coding work, while reserving cloud models for tasks that genuinely need them.
Agent 基建的三层爆发：电脑、运行时、技能方法论
从 Prime Agent、Cloudflare Computer、agent-skills、mattpocock/skills 和 Superpowers 看 Agent 基础设施正在如何从聊天框走向持久工作台、长任务运行时和工程方法论。
Tightening a dev VM that was quietly living without a firewall
What a routine security audit found on a small always-on dev VM: passwordless databases, no host firewall, over-privileged jobs, and how it got fixed.
Golden Rules for Cheaper, Safer LLM Agents
Notes on turning repeated AI workflows into scripts, skills, cron jobs, and guardrails so agents become cheaper, safer, and more useful over time.
Loop Engineering and Graph Engineering: Working With LLMs
A practical reflection on prompt, context, memory, loop, and graph engineering, and what each layer reveals about working with LLMs day to day.
When an Always Free server gets smaller
What Oracle's smaller Ampere A1 Always Free limits mean for a self-hosted ARM server, and how I redesigned mine around a 12 GB, 2-core budget.
The CI failures that moved every time you reran them
How a moving set of Rails CI flakes exposed shared download state, Selenium stale-node retries, and specs asserting before the browser settled.
Rebuilding a Small k3s Host Without Making It Precious
Notes from wiping an old single-node k3s setup and rebuilding it with a smaller, cleaner baseline for ingress, certificates, and persistent storage.
Don't Leave Good AI Workflows in Chat
A practical note on promoting repeated AI workflows into slash commands, scripts, and cron jobs, weighing judgment, cost, and safety trade-offs.
The Hard Part of AI PR Review Is Not Reading the Diff
A practical look at Alibaba OpenCodeReview, why diff-only AI review is not enough, and how to use AI reviewers without trusting them too much.
Durable codebase context for Hermes: GitNexus, LSP, AGENTS.md
How I gave Hermes a persistent local code graph, semantic diagnostics, and stable project instructions without turning agent memory into a stale symbol index.
Building an AI English Conversation Practice App
Notes on building an AI speaking practice app: STT, TTS, Gemini, ElevenLabs, latency, topic control, and useful feedback for oral English learners.
Running Hermes on a Cloud Machine as a Daily Routine Assistant
How I wired Hermes Agent on a remote Linux box into Telegram, email, calendar, error monitoring, GitHub, and a daily morning briefing routine.
Production Ruby Debugging: A Progressive Escalation Guide
A progressive escalation guide to debugging Ruby applications in Kubernetes production without restarting: rbspy, kubectl debug, signals, rbtrace, and gdb.
Tiny Puzzles for Testing and Debugging AI Agents
Small probe tasks for testing local AI agents: function calling, skill awareness, prompt-length pressure, cache behavior, and layer-by-layer debugging.
LM Studio Local Agent Runbook: Pi and OpenCode Step by Step
A step-by-step runbook for wiring LM Studio as a local OpenAI-compatible backend for Pi agents and OpenCode, with config examples and troubleshooting.
How an LLM Coding Agent Actually Builds Software
How LLM coding agents actually build software: the model, the tool loop, context management, precise patching, and verification against real tests.
Using LM Studio and Gemma as a Local Engine for Coding Agents
Using LM Studio and Gemma as a local OpenAI-compatible backend for coding agents: what broke, which settings mattered, and where it still falls short.
What Subagents Are Actually Good For in Codex
A field guide for AI-heavy early adopters on when subagents in Codex speed work up, when they slow it down, and how to use them without creating chaos.
Git Worktree for Parallel AI Development
How to use git worktree to run parallel AI coding sessions safely, with practical patterns for isolating gems, databases, ports, and runtime state.
MCP, Skills, Agents, Rulesets, and Hooks: A Practical Map
A practical introduction to the main building blocks of modern AI development, with diagrams that show how MCP, skills, agents, rulesets, and hooks work together.
Intelligent Applications and the Personal OS
The closing piece of a 12-part series on how software is shifting from apps you click through to intelligent systems and a personal OS that act for you.
I Let an AI Agent Write Code for a Week. Here's What Actually Worked.
A week of testing OpenHands, the autonomous AI coding agent, on real projects: what worked, what failed, and the setup that made it reliable.
How I Finally Got Sentry to Create GitHub Issues (After 3 Hours of Clicking Around)
How to fix Sentry's GitHub integration when it shows connected but issues won't create, plus why linking Sentry errors to GitHub issues is worth the setup.
Auditing USB-C Cable Quality from macOS Terminal
A macOS Terminal script that reads the I/O Kit registry to check negotiated wattage and voltage, so you can tell a fast USB-C cable from a slow one.
Stacked Git: a Patch-Stack Workflow on Top of Git
Stacked Git (StGit) separates the messy work of thinking through code from the clean history Git commits are for, using a patch-stack workflow.
Rails 8.1.1 + connection_pool 3.x: RedisCacheStore boot crash
A keyword-argument mismatch between Ruby 3.4 and connection_pool 3.x breaks RedisCacheStore at Rails 8.1.1 boot, fixed with a patch loaded before initializers run.
Choosing Your CSS Color Strategy: Static vs. Dynamic Variables
Static CSS color values vs. variables: why hard-coded colors bite you later, and how to structure colors so they stay maintainable.
From Shared Database to APIs: Taming the Performance Hit
How to cut the latency cost of moving from a shared database to an API-driven architecture, using caching, request aggregation, and async writes.
Closures, Blocks, and Iterators Across Six Languages
A working comparison of how Python, JavaScript, Ruby, PHP, Rust, and Go implement closures and lazy iteration, with a runnable counter and generator example for each.
JavaScript Engines in React Native: JSC vs. Hermes
React Native's default JS engine moved from JSC to Hermes for faster startup and lower memory use, and here is how to control which one your build uses.
Responsive Images: Serving the Right Size for Every Screen
Comparing srcset and picture, the native HTML approach to responsive images, against JavaScript solutions like jQuery Foundation Interchange.
Web Workers vs Service Workers: What Each One Is For
The difference between Web Workers and Service Workers: one offloads CPU work, the other intercepts network requests, and most apps eventually need both.
Ruby Ractor vs the GVL: When You Get Real Parallelism
How Ruby's Global VM Lock limits threaded CPU-bound code, and how Ractor bypasses it entirely by trading shared memory for message passing.
Is Your Endpoint I/O Bound or CPU Bound?
How to tell whether a slow endpoint is I/O bound or CPU bound, using system monitors, timing, and cProfile, and why the fix differs for each.
Debugging Exceptions: Break on Throw, Not Catch-All
Catch-all exception handlers hide the real bug. Debuggers that pause on every thrown exception, caught or not, find it faster than logging ever will.
Frontend Error Handling: Catching What try/catch Misses
Why Sentry and Bugsnag catch more JavaScript errors than custom try/catch logic, and how to close the gap with global handlers and rejection listeners.
Nokogiri vs libxml2: Debugging a Dynamic Library Conflict
Why Nokogiri XPath queries failed in Docker but not on an M1 Mac: a libxml2 version mismatch caused by indirect dynamic library loading in Ruby.
Timezone Bugs in Ruby: BST vs UTC in CI and Cron
How British Summer Time breaks Ruby tests, cron schedules, and cloud functions that assume UTC, and how to write timezone-aware specs instead.
Code Review Anti-Patterns That Waste the Most Time
The code review anti-patterns that slow teams down most: giant PRs, style-war comments, and reviews that skip the business context entirely.
编程方法论中的“矛盾”：权衡的艺术 (Ruby 示例)
本文用 Ruby 代码示例，剖析 DRY 与 AHA、YAGNI 与可扩展性设计、继承与组合等几对看似冲突的编程原则，说明该如何在实践中权衡取舍。
Debugging Transaction Deadlocks in Rails Tests: A Case Study
A Rails test suite hung intermittently until transactional tests, ActiveJob, and ActiveStorage turned out to be racing over the same locked rows.
Monkey-Patching in PHP vs Ruby: Why One Is an Anti-Pattern
Runtime method overrides work very differently in PHP and Ruby: one has no rollback mechanism, the other builds aliasing and scoping in from the start.
Why npm ci Runs prepare Scripts for Git Dependencies
npm ci is not the script-free install it looks like; it still runs prepare scripts for Git-based dependencies, which broke Husky in CI.
Handling Out-of-Order Updates in Async Rails Workflows
When async updates to the same record can finish in any order, the oldest write can silently overwrite the newest one; here is how to stop that.
Poetry vs uv: Choosing a Python Dependency Manager
Poetry and uv solve overlapping problems in Python dependency management; here is where each one wins, and how to use them together.
React Dev Proxy Options: package.json vs http-proxy-middleware
Two ways to avoid CORS errors in local React development: the package.json proxy field for one backend, or http-proxy-middleware for several.
Debugging in Ruby, Python, and PHP: byebug, pdb, and Xdebug
Compares Ruby's byebug, Python's pdb, and PHP's Xdebug side by side, with the core commands and a working example for each language.
Retry Strategies in Ruby: Exceptions vs. Conditional Checks
Comparing Ruby retry patterns, exception-based retry, conditional checks, and exponential backoff, and when each keeps performance acceptable.
API Design for Time Slot Replacement in Booking Systems
How to design a booking API that replaces regular time slots with alternatives, using server-side merging, cache versioning, and idempotency keys.
5 Parameter Code Smells and How to Refactor Them
Five parameter-related code smells, from boolean flags to long argument lists, and the refactor that turns each function into one that does one thing.
Building a Test Suite: Pyramid, Layers, and What to Skip
A framework for deciding how much to test: the testing pyramid, layered test responsibility by architecture tier, and what makes test data good.
Laravel Model: `save()` vs. `saveOrFail()`
When Laravel's save() silently returns false versus when saveOrFail() throws, and how to pair both with validation and database transactions.
Testing Non-ASCII Fonts with Prawn and PDF::Inspector
How to debug Unicode font rendering in Ruby's Prawn PDF library and work around PDF::Inspector's limits when testing non-ASCII characters like it.
Ruby readability vs performance: filter_map and a single loop
Ruby's filter_map cuts a multi-pass select/map/uniq chain to one pass, and a plain each loop goes further still when performance actually matters.
Android build and debug tips: jenv, ADB, and wireless debugging
Practical notes on Android build and debug workflow: managing Java versions with jenv, wireless ADB debugging, release signing, and logcat filtering.
Three-way merge in Neovim with Fugitive and Delta
Setting up Neovim, Fugitive, and Delta for three-way merges gives you a fast, keyboard-driven way to resolve Git conflicts without leaving the terminal.
Passing Sass Variables at Build Time with Webpack
How to inject a Sass variable, like a theme's primary color, at build time using a SASS_OPTIONS environment variable and a custom webpack additionalData function.
Upgrading React 17 to 18: Root API, Batching, and Test Fixes
What changes when you upgrade React 17 to 18: the new root API, automatic batching, Strict Mode's double-invoked effects, and test setup fixes.
Upgrading React 16 to 17: Dependencies, Jest, and ESLint Fixes
A practical walkthrough of upgrading a React 16 app to React 17, covering dependency updates, Jest and Enzyme test fixes, and new ESLint warnings.
adb shell pm: Managing Android Packages from the Command Line
How to use adb shell pm to list, install, uninstall, disable, and inspect Android packages, including removing bloatware without root.
Rufus-Scheduler vs Sidekiq-Scheduler for Ruby Jobs
Comparing rufus-scheduler's in-process polling loop with sidekiq-scheduler's Redis-backed queue to pick the right one for single-process vs distributed apps.
Rufus Scheduler: How Its Main Loop Actually Works
How rufus-scheduler's single-threaded main loop triggers, times out, and runs jobs, and where a timer-based rewrite would scale better.
Rack: The Interface Between Ruby Web Servers and Apps
Rack is the standard interface between Ruby web servers and frameworks, traced here through its middleware stack from request to response.
How to Debug React Applications
A rundown of debugging React apps: React Developer Tools in the browser, inspecting React Native, and running Jest tests under a real debugger.
Integer Overflow and Floating-Point Precision: Ruby vs JS
Why 0.1 + 0.2 doesn't equal 0.3, how Ruby's arbitrary-precision integers avoid overflow, and how JavaScript's BigInt does the same for large integers.
SCORM: What the E-Learning Standard Actually Does
SCORM is the packaging and tracking standard that lets e-learning content run and report progress consistently across different learning platforms.
Round Is Not Always What You Expect
Rounding the same float, 2.675, to two decimal places gives different answers in C#, PHP, Go, Python, JavaScript, and Ruby, for concrete reasons.
Floating-Point Precision in Ruby: Why 0.1 + 0.33 ≠ 0.43
Ruby's 0.1 + 0.33 does not equal 0.43 because of binary floating-point rounding, and Rational, BigDecimal, and tolerance comparisons each fix it differently.
A .irbrc Full of Ruby Debugging and Introspection Helpers
A collection of .irbrc helpers for tracing method calls, logging SQL queries, profiling code, and inspecting Ruby objects during Rails console debugging.
Shared Databases Between Microservices Are a Code Smell
When two services share one database, a schema change in either one can silently break the other, and here is how to split ownership with APIs or events.
Compress Images with FFmpeg
FFmpeg one-liners to scale and compress images: fixed dimensions, percentage scaling, and aspect-ratio-safe resizing.