How we measured a Gemini 3.1 Pro upgrade across a 25-call psychological analysis pipeline — comparing reasoning precision, verification quality, and tactic detection — and why the most valuable discovery was a silent wiring bug the upgrade forced us to find.
TanStack Start v1.132+ dropped Nitro as a default dependency, breaking Firebase App Hosting deploys. Here's how to restore the Nitro server runtime with the official compatibility plugin and get your SSR app running on Cloud Run.
This article explores how advanced mathematical concepts like Cauchy Convergence, Fisher Information, KL Divergence, and Wasserstein Distance can be applied to communication analysis to quantify deception, narrative drift, and strategic obfuscation, moving beyond traditional linguistic methods.
From agents and MCP to stateful DAGs and evals—why the AI Engineer is the critical bridge between raw LLMs and reliable enterprise systems.
IA is the invisible scaffolding behind great UX—structure, ontology, taxonomy, and choreography—and why it matters more in the age of AI.
How the Product Engineer archetype bridges code, design, and strategy—and why AI is accelerating this shift from outputs to outcomes.
We built a custom script that fetches shadcn components, rewrites their imports for a Deno workspace, wires up exports, and updates barrel files — then verified each component in Storybook before shipping to Astro and Next.js apps.
A practical guide to running Storybook 8 natively under Deno 2.6, covering the specific workarounds needed for config loading, workspace import resolution, and framework compatibility.
A single misconfiguration can send you down a rabbit hole of failed deployments. This is the story of the SSL redirect that breaks Fly.io health checks in Phoenix apps, and how to fix it.
Phoenix LiveView's DOM patching relies on stable element IDs. When IDs change between renders, hooks remount, state resets, and the DOM thrashes. This article examines a common pitfall with component ID generation and the correct solutions.
How to build multi-stage AI analysis pipelines that survive API timeouts, handle human review gates, and don't lose state when things go wrong — without heavyweight infrastructure.
Separating behavioural boundaries from user curation scopes to avoid silent no-ops, confusing semantics, and cross-context contamination in analysis systems.
Part 2 of the Code Quality series: the analytical missteps that revealed the architecture's true sophistication — the blooper reel, and the more important lesson.
How an AI-optimized monorepo architecture accelerates complex schema evolution across database, API, type definitions, and UI layers by an order of magnitude.
Part 1 of a two-part series documenting a code quality session that evolved from routine TypeScript fixes into deeper lessons about architecture and boundary layer patterns.
How our scaffolding CLI reduces entity creation from 15-20 minutes to 1 second by automating the 7-file boilerplate for the 4-layer architecture
Best practices and recommendations for implementing hybrid TOML + database pipelines to process large volumes of email threads with AI, covering chunking strategies, metadata management, and operational considerations.
How advanced TOML patterns — semantic registries, AI instruction blocks, and typed validation — enable constitutional governance and AI-assisted development.
MoonRepo already handles multi-app development beautifully with moon run :dev. But sometimes you want more control—explicit app selection, named prefixes, and production orchestration. Here's what we added and why.
Deno's modern runtime meets Node's mature ecosystem through polyglot Moonrepos. Learn how this architectural pattern transforms Deno from a niche runtime into a scalable production platform by letting Node build it and Deno run it.
drizzle-kit 0.31.x has a known bug that causes it to hang indefinitely. Here's how we diagnosed it, built a production-ready workaround, and what we learned from code review.
When a single moon run command builds 13 packages instead of 5, something's wrong. How we diagnosed over-broad Moon dependencies and achieved a 62% reduction in build time.
In a Deno-first monorepo, disabling type checking in dnt isn't a type safety gap — it's a deliberate optimisation. The real type checking happens at the Deno layer.
Most typecheck tasks either have no outputs or are cache-busted by listing the same file in both inputs and outputs. Here's how to audit and fix that.
How we designed an edge auth gateway pattern that removes all authentication logic from the frontend, implementing defence in depth while keeping the codebase simple.
Venice.ai offers the full spectrum of generative AI capabilities while keeping your data private through end-to-end encryption and decentralised processing.
How to structure a TypeScript monorepo so AI coding agents literally cannot make architectural mistakes
How to create a TypeScript architecture that catches AI agent mistakes in under one second through automated validation
A powerful build-time content processing library that transforms markdown and other content files into type-safe data collections with automatic TypeScript inference and framework integration.
A powerful combination for building type-safe, scalable database applications in TypeScript. Serverless PostgreSQL capabilities paired with a lightweight, performant ORM.
Building modern full-stack applications requires balancing performance, developer experience, and type safety. ORPC and Hono together provide a powerful combination that delivers all three.
A small, fast, and lightweight web framework designed specifically for modern edge computing environments. Built with performance and developer experience in mind.
This document outlines the real, practical, senior-engineer approach to building a robust, AI-driven monorepo. It details the 'why' of the stack, the 'how' of the structure, and the critical 'tricks' learned from experience.
A guide to integrating Storybook into a MoonRepo monorepo with Tailwind CSS v4. Covers component-driven development, documentation, and testing UI components in isolation.
Building professional software is a challenge of managing complexity. Building with Artificial Intelligence is a challenge of managing unpredictability. This document outlines a blueprint for a system of verifiable guarantees—a fortress of predictability and validation.
In Part 1, we covered our core 'data tools'—the Plan (TypeScript), the Inspection (Zod), and the Action (RPC). Now, let's look at the rest of the operation: the modern factory we use to build, run, and deliver our software to the world.
A guide to configuring Tailwind CSS Typography plugin in a monorepo with multiple frameworks. Covers plugin resolution, shared dependencies, and framework-specific setup.
A modern, type-safe authentication library designed specifically for full-stack TypeScript applications. Built with developer experience and security as top priorities.
All modern software runs on data. This is a simple story of how we plan, protect, and move that data. These three tools—TypeScript, Zod, and RPC—are the foundation for everything we build.
A consistent, efficient way to demo and adopt multiple webfonts across both Next.js and Astro apps without shipping every font by default. Covers opt-in loading strategies and shared utilities.
A comprehensive guide to MoonRepo's command-line interface for managing tasks across your monorepo. Covers essential commands for development, building, linting, and project management.
A cutting-edge build system specifically designed for managing monorepos with unprecedented efficiency and developer experience. Built from the ground up to handle complex, multi-language projects.