2026 W37: Codex Learns to Route Work (Agent Router)
A building-in-public account of turning Codex into a control hub for native subagents and external OpenCode, MiniMax, Big Pickle, and Antigravity workers.
Read entryDevlog
Building in public: Updates and progress.
W33: Orchestrating Performance, Astro Collections, and N-Squared Algorithms
A deep engineering chronicle on how I drastically reduced ArceApps build times through algorithmic tag optimization, related posts generation, and concurrent collection queries.
Read entry
W31: Bento Grids, SEO Agents, and Portfolio Architectural Redesign
Technical chronicle about ArceApps' modern portfolio redesign using Bento Grids, JSON-LD schemas, OpenCode automation, and Astro.
Read entry
W30: Editorial Home, Building in Public
Rebuilding the ArceApps landing page as an editorial cover: a typographic hero with the indie manifesto and four numbered sections (Devlog, Blog, Featured Work, CTA).
Read entry W28: Secure Architecture, Orchestration, and Performance in ArceApps
Chronicle of the latest fortnight in ArceApps' Building in Public. Details on XSS mitigation, testing with Vitest, and build optimizations.
Read entry
W26: SEO Audit, AI Agents, and the Evolution of the ArceApps Portfolio
A deep dive into my comprehensive SEO audit process, the evolution of comparative AI articles, and how I technically differentiate my portfolio from other projects like PuzzleHub.
Read entry
2026 W26: Connecting the dots of the puzzle
A week where two seemingly small features ended up exposing the deep architecture of PuzzleSuite: clickable navigation from the profile and retroactive persistence of XP and scores across all twelve games.
Read entry
2026 W26: The Rebellion of the Int? (Technical Deep Dive)
The B-side of this week: how to decide between nullable and non-nullable columns, why we wrote the RetroactiveStatsPopulator in Kotlin instead of SQL, and the guts of the Room migration for the twelve games of PuzzleSuite.
Read entry W25: Visual Architecture and Orchestrated Agents (ArceApps)
Complete Homepage redesign with an interactive Bento Grid, Content-Visibility enhancements, and deep advancements in AI agent documentation.
Read entry W23: Enhancing Testing and Documentation Sync (ArceApps)
Engineering chronicle on adding comprehensive tests for search functionality and syncing visual hierarchy documentation with CSS in the ArceApps portfolio.
Read entry W22: Socratic Agents and OpenSpec Expansion (ArceApps)
Engineering chronicle detailing the release of the Socratic Agents blog series and the introduction of OpenSpec for mobile development in the ArceApps ecosystem.
Read entry W20: Bilingual Refactoring, Optimization, and Security (ArceApps)
An engineering chronicle detailing architectural improvements, performance optimizations, XSS security mitigations, and enriched unit testing at ArceApps.
Read entry W18: Hardening the Portfolio, DOM XSS Mitigation, and Visual Hierarchy (ArceApps)
A deep dive into security patching in Astro, refining responsive visual hierarchy for Markdown content, and chronicling the genesis of mydevbot.
Read entry W15: Architecture and Security Evolution at ArceApps (ArceApps)
A deep dive into how we reinforced security by mitigating XSS vulnerabilities, optimized the visual hierarchy of our media, and upgraded our AI agents to enhance brand consistency across the ArceApps portfolio.
Read entry
The Maturity of mydevbot - CI/CD, eGPUs and the Future of Mobile Development (mydevbot)
The final phase. How mydevbot deploys itself using GitHub Actions and Watchtower, mobile programming with VS Code Server and the preparation of the UM890 Pro for extreme local AI via OCuLink.
Read entry
Teaching mydevbot to Program - GitHub Skills and Cron Tasks (mydevbot)
The second phase of mydevbot. How to integrate the GitHub API using Gemini's Function Calling capabilities and set up scheduled tasks with APScheduler to receive daily summaries.
Read entry
The Birth of mydevbot and the Odyssey of the Perfect Hardware (mydevbot)
How the need to control my development ecosystem from Telegram arises, the analysis of the AIPAL repository, the disappointment with the Synology DS212+ and the final decision for a Mini PC with the native Gemini SDK.
Read entry 2026 W07: The Assembly Line (PuzzleHub)
How we went from craftsmanship to industrial production: 11 interactive tutorials in 7 days. A technical chronicle on scaling, technical debt, and the art of teaching geometry without using words.
Read entry 2026 W06: Teacher Mode (PuzzleHub)
We redesigned the onboarding system. No more static texts. We implemented an interactive 'Teacher Mode' that guides you step by step, detects your confusion, and celebrates your small victories.
Read entry 2026 W05: The Quest for Mathematical Truth (PuzzleHub)
A 3000-word chronicle on how we tried to add two 'simple' games and ended up fighting NP-Complete problems. The story of how a simple light board forced us to use industrial SAT Solvers and how integer division almost destroyed our mental health.
Read entry
2026 W04: The Silent Referee (PuzzleHub)
A technical and philosophical dive into eliminating UI bureaucracy. How we moved from asking the user to fill out forms to creating a Reactive Auto-Verification system, and how we optimized O(N) graph cycle detection to run at 60fps.
Read entry 2026 W03: The Safety Net (PuzzleHub)
The detailed technical story behind the implementation of the Undo/Redo system in PuzzleHub: design patterns, concurrency problems with coroutines, efficient memory management, polymorphic serialization, and the search for the perfect UX on Android.
Read entry
2026 W02: Generics, Canvas, and the Quest for Absolute Efficiency (PuzzleHub)
A chronicle of extremes: eliminating 400 lines of repeated code with generic architectures and redrawing every pixel by hand to save performance on mid-range devices.
Read entry
2026 W01: House Cleaning and Solid Foundations (PuzzleHub)
Starting the year by bringing order to chaos. A technical chronicle of how we dismantled a 600-line ViewModel, unified time logic with Flows, and improved Slitherlink's invisible accessibility.
Read entry Research: Clawdbot and Local-First Architecture (Clawdbot)
Research diary on Clawdbot. Beyond the tutorial, here I analyze the decentralized architecture and Gemini 3.0 integration challenges.
Read entry
Refining the Experience: Performance and Invisible Details (ArceApps)
Confessions of a week optimizing ArceApps. The battle against CSS overlays, lazy image loading, and why I refactored search.
Read entry
2025 W53: Closing the Year with Expert Level (PuzzleHub)
Unification of difficulty systems across the suite and the arrival of Expert mode to bid farewell to 2025.
Read entry
2025 W52: House Cleaning and Solid Foundations (PuzzleHub)
Massive refactoring of game states and visual improvements in Hashi before closing the year.
Read entry
2025 W51: Breaking the Language Barrier (PuzzleHub)
A massive internationalization sprint to eliminate hardcoded text and prepare PuzzleSuite for the world.
Read entry
2025 W50: Perfecting The End (PuzzleHub)
Improving the puzzle completion experience: more useful, standardized dialogs with the option to admire your work.
Read entry
2025 W49: Achievements Unlocked (PuzzleHub)
Minesweeper and Galaxies join the party, completing the achievement system for all 10 games.
Read entry
2025 W48: Invisible Foundations (PuzzleHub)
Preparing the ground for feature parity. Backend work, Room migrations, and why Minesweeper is different from everything else.
Read entry 2025 W47: The White Lie and the Redesign (PuzzleHub)
How we redesigned the app's main entrance and discovered we had been lying to our users about their stats.
Read entry
2025 W46: The Broken Window Theory (PuzzleHub)
Why we decided to halt development of new features to fix a pixel 6dp out of place, and the hunt for the geometric bug in Hashi.
Read entry
2025 W45: Dominating Uncertainty (PuzzleHub)
From theory to practice: How we implemented Dominosa in a week and learned that true randomness is an enemy.
Read entry
2025 W44: The Ninth Passenger (PuzzleHub)
Just when you thought the catalog was closed, Dominosa arrives to challenge our architecture. A story about feature creep and modular design.
Read entry
2025 W43: The Cambrian Explosion (PuzzleHub)
Doubling the catalog in seven days. From 4 games to 8. Code reuse reaches its peak, but technical debt starts accumulating interest.
Read entry
2025 W42: Forbidden Geometry (PuzzleHub)
Slitherlink and Galaxies break our grid. Introducing vector math, fuzzy hit zones, and background thread generation.
Read entry
2025 W41: Multiply by Three (PuzzleHub)
From one game to three in record time. How the right abstraction and Kotlin Generics allowed us to scale without copying code.
Read entry
2025 W40: The First Pixel (PuzzleHub)
The fear of the blank page, the critical decision between Canvas and Compose, and why we chose an architecture that slowed us down for three days to speed us up for the rest of the year.
Read entry
The Indie Tech Stack: Reflections on Architecture and Autonomy (ArceApps)
An incredibly detailed review of the ArceApps system architecture, custom AI agents, and the philosophy of solo software engineering.
Read entry
Automated QA as an Indie: Testing the ArceApps Ecosystem (ArceApps)
A comprehensive analysis of how I built an enterprise-grade automated testing pipeline using Playwright and Vitest for a single-developer portfolio.
Read entry
Mastering Responsive Visual Hierarchy: The Definitive Indie Guide (ArceApps)
A massive deep dive into how I completely rebuilt the ArceApps UI architecture using Tailwind CSS v4, Astro, and fluid typography principles.
Read entry