vela-commerce
yasinatesim/vela-commerce
Open-source, AI-agent-driven e-commerce reference platform built with a disciplined Claude Code agent ecosystem.
Instalação
npx convez-ai yasinatesim/vela-commerce1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
What is Sklent?
Sklent is a Claude Code agent bootstrap. Agents, skills, references, and hooks turn project rules into mechanically enforced constraints.
The agent ecosystem lives under examples/e-commerce/.claude/. The repo ships one worked example: a full-stack e-commerce platform under examples/e-commerce/.
sklent/
├── examples/
│ └── e-commerce/
│ ├── .claude/ ← agents, skills, references, hooks
│ │ ├── agents/ code reviewers, BRAID solver, auditors
│ │ ├── skills/ braid-plan, ship-pr, coverage-gate, security-pentest
│ │ ├── references/ coding, backend, frontend, security standards
│ │ └── hooks/ shell gates: no aliased imports, no protected-branch commits
│ └── CLAUDE.md project rules loaded every session
Full project reference: Product Requirements Document
The BRAID reasoning model
An agent's worst failure is error compounding. A mistake in one step feeds the next. A flat to-do list never defines what happens on failure.
BRAID (Bounded Reasoning for Autonomous Inference and Decisions, arXiv 2512.15959) structures a task as a graph of four node types: Constraint, Fact, Step, Check. Every Check has two edges, Pass and Fail. Fail loops back to an earlier Step. See examples/e-commerce/.claude/references/braid-mental-model.md.
The agent ecosystem
Everything in examples/e-commerce/.claude/:
agents/--wtf-code-reviewerdispatches a diff towtf-go,wtf-js-react,wtf-security,wtf-ux-playwrightin parallel. Alsobraid-solver,constants-guard,issue-auditor.skills/--braid-plan,spec-driven-development,coverage-gate,playwright-snapshot,ship-pr,issue-create,security-pentest(web/api/network),intended-vs-implemented.references/-- language-agnostic coding standards, backend/frontend/security standards, git-flow, and the BRAID mental model.hooks/-- shell scripts that block aliased Go imports, direct commits to protected branches, agent-initiated merges, and 2-line comments. Runs CI-mirror verify before commit.
The example: Vela Commerce
A brandless e-commerce platform that runs with one docker compose up. Go 1.25 API (Gin, GORM, Postgres), Next.js storefront and admin (TR/EN), Iyzico 3D Secure sandbox, ChromaDB RAG product copy, GIB e-Arsiv invoice proxy, marketplace skeletons. See examples/e-commerce/.
cd examples/e-commerce && cp .env.example .env && docker compose up --build
Where the ideas come from
Adapted from public work with attribution in each agent and skill. The project concept and BRAID integration approach are documented in the project manifesto:
- Claude Code: Subagents
- Claude Code: Agent Skills
- Claude Code: Hooks
- Anthropic Cookbook
- mukul975/Anthropic-Cybersecurity-Skills (inspiration for
security-pentest) - phuryn/pm-skills (inspiration for
intended-vs-implemented) - BRAID arXiv 2512.15959
License
MIT. See LICENSE.