Ejecuta cualquier Skill en Manus
con un clic

Ejecuta cualquier Skill en Manus con un clic

code-auditor

Estrellas2

Forks1

Actualizado9 de marzo de 2026, 15:29

Comprehensive codebase analysis covering architecture, code quality, security, performance, and maintainability. Activates when user says "audit", "review the codebase", "check for issues", "technical debt", "security audit", "code quality check".

Instalación

Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.

Ejecutar en Manus

Fuente

IvyNotFound

IvyNotFound/KanbAgent

Abrir repositorio de GitHub Ver repositorios del creador

Descarga

Ejecutar en Manus

Ocupaciones relacionadasSOC

Basado en la clasificación ocupacional SOC

Analistas de seguridad de la informaciónOcupaciones informáticas y matemáticas·SOC 15-1212

SKILL.md

readonly

name	code-auditor
description	Comprehensive codebase analysis covering architecture, code quality, security, performance, and maintainability. Activates when user says "audit", "review the codebase", "check for issues", "technical debt", "security audit", "code quality check".

Code Auditor Skill

Full audit of the KanbAgent codebase across 5 dimensions, adapted to the Electron + Vue 3 + TypeScript + SQLite stack.

When to Use

User says "audit the code", "check for issues", "review quality"
Before a major release or refactor
After a long development sprint
When onboarding a new contributor

Audit Dimensions

1. 🏗️ Architecture

Is the Electron main/renderer separation clean? No business logic in renderer?
Are IPC channels well-typed and documented?
Is the Vue component hierarchy logical? (smart vs dumb components)
Are Pinia stores scoped correctly? No God stores?
Is the SQLite layer properly abstracted (repository pattern or equivalent)?

2. 🔒 Security

Is all IPC input validated in the main process? (never trust renderer)
Are SQLite queries parameterized? (no string interpolation in queries)
Is contextIsolation enabled in Electron? Is nodeIntegration disabled in renderer?
Are there any exposed sensitive data in IPC channels?
Check preload.ts for unnecessary API surface exposure

3. ⚡ Performance

Any synchronous operations blocking the main process?
Large SQLite queries without pagination or indexes?
Vue components re-rendering unnecessarily? (missing computed, shallowRef)
Any memory leaks in IPC listeners? (listeners not cleaned up on component unmount)

4. 🧹 Code Quality

TypeScript strict mode compliance? No any types?
Vue 3 Composition API patterns consistent? (<script setup> everywhere)
Dead code, commented-out code, unused imports?
Functions too long (>50 lines)? Complex conditionals that need extraction?
Consistent error handling across IPC calls?

5. 🧪 Maintainability

Are components testable in isolation?
Is there a clear naming convention for IPC channels?
Are DB migrations versioned and reversible?
Is the CLAUDE.md up to date with current architecture?

Output Format

## Code Audit — KanbAgent
Date: [date]

### Summary
[2-3 sentence overall assessment]

### Critical Issues 🔴
- [issue] — [file/location] — [recommended fix]

### Important Issues 🟡
- [issue] — [file/location] — [recommended fix]

### Minor Issues 🟢
- [issue] — [file/location]

### Strengths ✅
- [what's working well]

### Recommended Next Steps
1. [priority action]
2. ...

Notes

Focus on actual issues found, not hypothetical ones
Be specific: include file names and line numbers when possible
Don't flag style preferences as issues unless they cause real problems

Más de este repositorio

mismo repositorio

agent-session-starting

IvyNotFound/KanbAgent

Start an agent session correctly in KanbAgent. Activates when an agent needs to begin work, when user says "start a session", "launch agent X", "begin session", or when an agent role is invoked and no active session exists. Ensures correct startup sequence: dbstart → read tasks → begin work.

2026-04-052

agent-log-writing

IvyNotFound/KanbAgent

Write agent_logs entries correctly in KanbAgent. Activates when an agent needs to log an action, when completing a task, starting a session, or when agent_logs entries are missing. Triggers on "log this", "insert agent_log", "add a log", or any task_started/task_done/session_end event.

2026-04-012

adr-writing

IvyNotFound/KanbAgent

Write an Architecture Decision Record (ADR) for KanbAgent following the project's existing format. Activates when a structural decision needs to be documented, when agent arch completes a design decision, or when user says "ADR", "architecture decision", "document this choice".

2026-03-102

ipc-handler-creating

IvyNotFound/KanbAgent

Create a new typed IPC handler in KanbAgent following the full stack pattern. Activates when user or agent needs to add a new IPC channel, expose a new API to the renderer, or wire a new feature across main/preload/renderer. Triggers on "add IPC", "new handler", "expose to renderer", "new IPC channel", "create a handler".

2026-03-102

migration-creating

IvyNotFound/KanbAgent

Create a versioned SQLite migration for KanbAgent. Activates when the DB schema needs to change, when user says "add a column", "create a table", "migration", "alter schema", "add a column", "modify the schema". Follows the existing migration pattern with idempotency and better-sqlite3 compatibility.

2026-03-102

release-checklist

IvyNotFound/KanbAgent

Execute a release in KanbAgent following the full devops + review protocol. Activates when agent devops receives a release ticket, when user says "release", "publish version", "bump version", "npm run release". Covers prerequisites, version bump, CI wait, GitHub Release publication.

2026-03-102

Code Auditor Skill

Full audit of the KanbAgent codebase across 5 dimensions, adapted to the Electron + Vue 3 + TypeScript + SQLite stack.

When to Use

User says "audit the code", "check for issues", "review quality"

Before a major release or refactor

After a long development sprint

When onboarding a new contributor

Audit Dimensions

1. 🏗️ Architecture

Is the Electron main/renderer separation clean? No business logic in renderer?

Are IPC channels well-typed and documented?

Is the Vue component hierarchy logical? (smart vs dumb components)

Are Pinia stores scoped correctly? No God stores?

Is the SQLite layer properly abstracted (repository pattern or equivalent)?

2. 🔒 Security

Is all IPC input validated in the main process? (never trust renderer)

Are SQLite queries parameterized? (no string interpolation in queries)

Is contextIsolation enabled in Electron? Is nodeIntegration disabled in renderer?

Are there any exposed sensitive data in IPC channels?

Check preload.ts for unnecessary API surface exposure

3. ⚡ Performance

Any synchronous operations blocking the main process?

Large SQLite queries without pagination or indexes?

Vue components re-rendering unnecessarily? (missing computed, shallowRef)

Any memory leaks in IPC listeners? (listeners not cleaned up on component unmount)

4. 🧹 Code Quality

TypeScript strict mode compliance? No any types?

Vue 3 Composition API patterns consistent? (<script setup> everywhere)

Dead code, commented-out code, unused imports?

Functions too long (>50 lines)? Complex conditionals that need extraction?

Consistent error handling across IPC calls?

5. 🧪 Maintainability

Are components testable in isolation?

Is there a clear naming convention for IPC channels?

Are DB migrations versioned and reversible?

Is the CLAUDE.md up to date with current architecture?

Output Format

## Code Audit — KanbAgent Date: [date] ### Summary [2-3 sentence overall assessment] ### Critical Issues 🔴 - [issue] — [file/location] — [recommended fix] ### Important Issues 🟡 - [issue] — [file/location] — [recommended fix] ### Minor Issues 🟢 - [issue] — [file/location] ### Strengths ✅ - [what's working well] ### Recommended Next Steps 1. [priority action] 2. ...

Notes

Focus on actual issues found, not hypothetical ones

Be specific: include file names and line numbers when possible

Don't flag style preferences as issues unless they cause real problems