vibeshield

A security and architecture review assistant for the VibeDev development cycle. It looks for risks when adding login, stored data, third-party APIs, production deployment, or dependencies, then explains them in technical or plain language.

In plain words
What is it for?
Use it to review security and architecture risks around authentication, data storage, external services, deployment, and new packages.
Why use it?
It helps identify problems before they become security incidents or costly design changes. It presents findings and choices but does not edit code or decide for the user.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/4pixeltechbr/vibedev/vibeshield
Any agent
npx skills add 4pixeltechBR/VibeDev --skill vibeshield
Clone the repo
git clone --depth 1 https://github.com/4pixeltechBR/VibeDev

Made for: Claude Code, Codex.

Per session 68 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,195 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00068 $0.02195
Opus 5 $0.00034 $0.01097
Sonnet 5 $0.00014 $0.00439
Haiku 4.5 $0.00007 $0.00219

Measured 2d ago against content hash 4e15a6571b6c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

vibeshield scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

vibeshield/SKILL.md · 147 lines

How it starts

The opening of the file, as written. The whole thing — 147 lines — stays where its author put it; the contents beside it link to each section on GitHub.

VibeShield

Você é o Mentor Tradutor de segurança do projeto. Seu trabalho é enxergar o que pode dar errado antes de virar problema, traduzir isso pra linguagem que o usuário entende, e devolver o controle pra ele decidir.

Para o engenheiro: usa categorias técnicas (C1-C8), gatilhos (G#) e verdicts (OK/REVISAR/BLOQUEAR). Tudo documentado em references/handoff-vibedev.md.

Para o leigo (campo modo_usuario: leigo no PROJECT_STATE.md): usa os templates conversacionais em references/formato-conversa-leigo.md, traduzindo riscos via references/glossario-leigo.md. Mostra progresso com o painel em references/estado-visual.md.

Princípios (não negociáveis)

  1. Não tem mãos. Analisa, categoriza, recomenda. Não edita código, não roda scanner, não modifica estado fora da seção da sub-tarefa ativa.
  2. Não decide pelo usuário. Devolve opções; usuário escolhe.
  3. Bloqueio é bloqueio. Verdict BLOQUEAR não pode ser "aceito por inércia". Se for pulado, é decisão Tipo 1 com motivo registrado.
  4. Estado mora em PROJECT_STATE.md. Não cria arquivo próprio. Lê, age, escreve timestamped na sub-tarefa ativa.
  5. Tom é mentor paciente, não auditor categórico. Aberturas reconhecem o que o usuário tá fazendo. Encerramentos devolvem controle.
  6. Regra dos 3 minutos. Toda conversa cabe em 3 minutos de leitura. Se passar, divide e pergunta por onde continuar.

Quando você entra em ação

Você é chamada pela VibeDev quando um dos 7 gatilhos (G1-G7) bate. Não invente caminho próprio. Em resumo:

Gatilho O que disparou Momento
G1 Sub-tarefa toca autenticação pré-build
G2 Sub-tarefa toca persistência de dados pré-build
G3 Sub-tarefa toca integração com API de terceiro pré-build
G4 Sub-tarefa toca exposição pública / deploy pré-build
G5 Dependência nova entra no projeto durante check
G6 Sub-tarefa reprovou 2x no /vd-check modo debug
G7 Usuário chamou /vibeshield-audit explicitamente qualquer hora

Read the full file on GitHub · 147 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 147 lines · 0 tokens per session scan A 4e15a6571b6c

Subscribe to this mod's changes

vibeshield is a skill published in the GitHub repository 4pixeltechBR/VibeDev (16 stars, last pushed 1mo ago), licensed MIT. It adds 68 tokens to every session and 2,195 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

debug-go

Systematic 5-step debugging flow for Go 1.22+ services. Load when a test fails, a goroutine leaks, a downstream provider hangs, errors lose context, or production logs are unhelpful. Forces layer isolation (handler vs service vs repo vs provider) and runs the 5 most common Go antipattern greps before any code change …

yerdaulet-damir/vibe-coding-rules · 91 tokens

add-provider

Checklist for adding a new AI provider (image, video, text, audio) that satisfies all 5 integration principles — ACL, bulkhead, idempotency, observability, and contract test. Load when integrating any new external AI API. Prevents the most common mistake of pasting httpx calls directly into a service.

yerdaulet-damir/vibe-coding-rules · 69 tokens

debug-frontend

Systematic 5-step debugging flow for Next.js 15 + React 19 + TypeScript apps. Load when a UI bug is reported, hydration error appears, Server Action returns wrong data, or cache invalidation fails. Forces layer isolation (server vs client, action vs component, cache vs render) before touching code — prevents "fix…

yerdaulet-damir/vibe-coding-rules · 80 tokens

new-feature-go

Pre-flight checklist for adding a new feature to a Go 1.22+ service. Load when creating a new endpoint, internal package, external integration, or background worker. Forces "accept interfaces, return structs", context.Context propagation, error wrapping, and bulkhead-per-provider from line one — prevents the most…

yerdaulet-damir/vibe-coding-rules · 91 tokens

new-feature-nextjs

Pre-flight checklist for adding a new feature to a Next.js 15 + TypeScript app. Load when creating a new page, route, server action, or feature module. Forces feature-driven colocation, RSC-first thinking, and the typed cache-tag pattern from line one — prevents the most common AI failures (1500-line page.tsx, 'use…

yerdaulet-damir/vibe-coding-rules · 90 tokens

debug-backend

Systematic 5-step backend debugging flow for AI-coded FastAPI apps. Load this skill when a bug is reported, a test fails, or unexpected behavior appears in any backend layer. Forces layer isolation before touching code — prevents the "touch 8 files and make it worse" pattern.

yerdaulet-damir/vibe-coding-rules · 62 tokens