Search
Posts
Live Data Backbone: Cache Layers, Circuit Breakers, and SSE for Broadcast
Wiring MLB Stats API, GUMBO feeds, and Statcast into a fault-tolerant data layer with TTL caching, circuit breakers, and Server-Sent Events.
ReadZero to CI: Full-Stack Dashboard in One Session
11 commits from empty repo to green CI. Building the Braves Booth Intelligence dashboard with opinionated stack choices and a CAD agent side quest.
ReadEngine to Product: Three Interfaces, One Codebase
Building CLI, REST API, and frontend wizard interfaces on top of a comparison engine. The jump from library to usable product in cad-dxf-agent.
ReadGolden Tests, Fuzz Testing, and a Nasty Fixture Taxonomy for DXF Revisions
Building a test corpus with clean/nasty fixture taxonomy, golden file regression tests, and fuzz testing for a DXF comparison engine.
ReadBuilding a Deterministic DXF Comparison Engine in One Day
Four PRs, 5000+ lines, 1875 tests. How to compare two versions of an engineering drawing when the coordinates might be slightly different.
ReadThe Silent Killer in Your Web App: How Bare catch {} Blocks Hide Failures from Everyone
A real-world case study from a live AI-powered CAD DXF editor: how two bare catch {} blocks swallowed every render failure silently, leaving users staring at a placeholder forever — and the full fix, PR review, and deployment cycle to surface the truth.
ReadCAD Agent v0.2.0: Planner Self-Correction, Hustle Auth Consolidation, and Moat Dual-Mode Gateway
Eight commits ship a validation feedback loop for the CAD planner, a single-cookie auth consolidation kills an XSS vector in Hustle, and Moat gets a dual-mode MCP/Web3 gateway.
ReadCAD Agent Perf Audit and git-with-intent SPIFFE Identity Analysis
A performance gap audit for the CAD agent and a SPIFFE/A2A identity analysis for git-with-intent — two planning sessions, no shipping.
ReadFebruary 2026 State of Affairs: 255 Commits, 12 Projects, and What I Shipped
A candid look at 18 days of engineering output: 255+ commits across 12 projects, from blockchain infrastructure to CAD agents to auth debugging.
ReadBuilding Moat: Auth, On-Chain Receipts, and 117 Integration Tests in One Week
How I built a policy-enforced execution layer for AI agents with HTTP proxy allowlists, IRSB on-chain receipts, and 117 integration tests — plus the CI saga that took 6 commits to fix.
Read