ForgeWard
2026
A model-agnostic, security-first software delivery harness that turns specialist model roles into a gated SDLC with human approvals and inspectable evidence.
Job Application Batch Builder
2026
An evidence-first Codex plugin and reusable skill that researches current roles and builds truthful, role-specific application packages without quietly inventing a better candidate.
Central Deploy Manager
2026
The deployment plumbing behind this site and its smaller siblings: signed webhooks, health-checked Docker rollouts, and Caddy routing on one VPS.
Aquarium
2026
A small 3D aquarium for the browser, built with React and Three.js and deployed like a real app because apparently I cannot leave anything simple.
Bird of the Day
2026
A tiny daily bird site powered by recent eBird observations, a small Express API, and an unreasonable amount of affection for birds.
CentraID
2026
A six-person capstone connecting a NestJS/PostgreSQL backend, an Expo mobile app, and a classroom reader service for verified check-ins.
React Native Cloud Prototypes
2025
Two Expo/Firebase prototypes: one for vehicle access approvals and one for QR-based classroom attendance.
Wurmkickflip
2026
An extremely serious physics experiment about teaching a worm to ride a skateboard. The physics works; the worm remains a work in progress.
2026
Job Application Batch Builder
An evidence-first Codex plugin and reusable skill that researches current roles and builds truthful, role-specific application packages without quietly inventing a better candidate.
My notes
I built this in the middle of my own job search, after discovering that tailoring applications is repetitive but automating it carelessly is a spectacular way to produce polished nonsense.
The skill starts by separating evidence from assumptions. It can inspect candidate-authorized resumes, transcripts, repositories, GitHub profiles, constraints, and current job postings, then use that map to prepare role-specific materials. Claims still have to trace back to something real, related projects can be consolidated without becoming imaginary employment, and application submission stays outside the workflow unless somebody authorizes it separately.
I also took it through the less glamorous parts of publishing a real plugin: a public repository scrubbed of candidate data, reusable scripts and references, a small product site, brand assets, and plain-language privacy, terms, and support pages. Building a job-search tool while looking for a job is either good systems thinking or an extremely elaborate form of procrastination. I am choosing the first interpretation.
Job Application Batch Builder is a reusable Codex plugin and skill for producing evidence-based application batches from candidate-authorized sources.
The workflow builds a structured evidence map, researches and ranks current roles, and generates ATS-readable role-specific resumes, cover letters, application-answer guides, and supporting DOCX/PDF packages. It audits claims, privacy, file structure, and rendered document output before handoff. The skill explicitly avoids inventing employers, dates, metrics, qualifications, or skills, and it does not submit applications or contact employers without separate authorization.
The public repository contains the plugin manifest, reusable skill instructions, reference material, local privacy-audit utilities, and brand assets. It intentionally excludes resumes, transcripts, generated applications, contact details, credentials, and candidate-specific material. A live public site documents the product and provides its privacy policy, terms, and support routes.