Skip to content

About

Opinionated operating-profile skill for AI coding agents — locked stack, strict workflow, mandatory pre-push gate. Installable via npx skills.

Resources

Stars

2 stars

Watchers

0 watching

Forks

Repository files navigation

tdvorak-fullstack

An opinionated operating-profile skill for AI coding agents: one SKILL.md plus reference docs that turn any agent harness into a disciplined full-stack engineer — a locked stack, a strict workflow order, hard rules, and a mandatory pre-push gate.

Used daily across real production repos. Run it stock, fork it, or let your agent reshape it to your stack and identity.

Install

npx skills add Dvorinka/tdvorak-fullstack

or manually:

git clone https://github.com/Dvorinka/tdvorak-fullstack
cp -r tdvorak-fullstack ~/.agents/skills/   # or your harness's skills dir

Works with Devin, Claude Code, Codex, Cursor, Windsurf, OpenCode, Gemini, Grok, Copilot — any harness that reads SKILL.md skills or plain rule files.

Customize — three modes

Mode Command Who answers
Agent + user (recommended) point your agent at SETUP.md agent inspects your env, asks you a few questions, generates
User only ./setup.sh you, via interactive questionnaire
Agent only cp customize.example.yml customize.yml, then ./setup.sh --config customize.yml --install --yes agent fills the answers itself
Stock ./setup.sh --stock nobody — profile ships as-is

The output is a complete skill directory under ./dist/<skill-name>/ (generated, gitignored); --install copies it into every detected agent skills directory on the machine.

What the profile enforces

  • Locked stack — defaults below; setup.sh rewrites the table to your stack
  • Workflow order — read the right reference doc before the kind of work it governs
  • Hard rules — structured tools for structured files, no skipped-test-as-green, task-runner-first, visual verification of UI changes
  • Mandatory pre-push gate — build, tests, lint/typecheck, smoke test; a failing app is a hard block, not a warning
  • Docs freshness gate — docs, changelog, OpenAPI, env docs, CLI help updated in the same diff
  • PR discipline — feature branch, authored commits, merge commits (no squash)
  • Persistent memory — durable facts live in a git repo (~/agent-memory), not session context
  • Token discipline — smallest correct diff; compression skills on demand

Default stack (all linked — swap freely)

Layer Default
Frontend React 18 + Vite + TS strict + Tailwind + shadcn/ui
Backend Go + Gin, zap
DB PostgreSQL + sqlc + goose
Cache DragonflyDB primary · Redis second option
API contract OpenAPI → generated clients
Infra Docker Compose
Desktop Wails / Tauri
Mobile React Native + Expo
E2E Playwright · Maestro · tauri-driver
Load k6
Lint/format oxlint + oxfmt + anti-slop plugin · golangci-lint
Hooks lefthook
Task runner just
Shell env direnv
API collections Bruno
Worktrees worktrunk
Agent orchestration t3code
Local CI act · actionlint · shellcheck
HTTP debug mitmproxy
Security trivy · ZAP baseline · gitleaks · govulncheck · security-audit-skill
Search/data ripgrep · fd · jq · yq · TOON · sqlglot
DB tests testcontainers-go
Maps MapCN
Design Boneyard · shadcn/ui
Auth Better Auth
Billing stripe-kv-sync skill + t3dotgg/stripe-recommendations guidance
Code health desloppify skill — post-build structure/maintainability/debt scan
Memory Agent Memory Repo spec

Companion agent CLIs (optional, by the same author)

Tool Use
docsync docs ↔ code drift check before PRs
envdiff "works on my machine" environment diffs
ciwatch CI wait/failure triage + local preflight
repolint repo hygiene audit before publishing

go install github.com/Dvorinka/<name>/cmd/<name>@latest — all emit --json.

Layout

SKILL.md                 the operating profile (agent reads this)
references/              deep docs, loaded on demand by the workflow
  architecture.md        backend/DB/API contract rules, TS evidence rules
  ui-standards.md        design bar, component selection, review skills
  toolchain.md           lint/format/test/CI machinery
  maps.md                map component policy
  terminal-data.md       search/JSON/pipeline discipline
  agent-ops.md           parallel dispatch, worktrees, remote machines
  t3code.md              T3 Code orchestration surface and usage guide
  ship-verify.md         deploy-and-verify discipline
  go-dbtest.md           DB-backed Go test recipe
  release-train.md       release mechanics conventions
  upstream-port-sweep.md fork sync loop
  handoff.md             session-close handoff convention
setup.sh                 personalization generator
SETUP.md                 agent playbook (the recommended flow)
customize.example.yml    answer-file template

Philosophy

Opinionated on purpose. Every rule in SKILL.md either prevents a failure that actually happened, or enforces a discipline that pays for itself daily. Change any of it — that's what --config is for — but change it deliberately.

License

MIT — see LICENSE.

About

Opinionated operating-profile skill for AI coding agents — locked stack, strict workflow, mandatory pre-push gate. Installable via npx skills.

Resources

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages