j4rk0r

Revisa el diff de la rama actual frente a develop en proyectos Drupal 11 siguiendo la metodología Codex (lógica de negocio, edge cases de hooks/queries, seguridad, performance, completitud). Genera un informe .md en la carpeta del IDE detectado (.antigravity/, .cursor/, .vscode/ o docs/) con hallazgos por severidad y soluciones accionables. Usar cuando el usuario pida "Revisión diff develop", "revisión diff develop", "diff develop", "revisar diff", "codex diff" o expresiones similares con intención de auditar cambios contra develop. Triggers: diff develop, codex diff, revisión diff, lint diff develop, auditar diff.

Revisa pull requests en proyectos Drupal 11 (u otro) siguiendo la metodología Codex (lógica de negocio, edge cases de hooks/queries, seguridad, performance, completitud). Genera un informe .md en la carpeta del IDE detectado (.antigravity/, .cursor/, .vscode/ o docs/) con hallazgos por severidad y soluciones accionables. Usar cuando el usuario pida "revisión Codex", "revisión de PR", "revisar PR", "revisar PR #N", "revisión tipo Codex" o indique un número de PR con ramas (ej. develop ← feature/alejandro). Triggers: revisión PR, revisar PR, codex PR, revisión Codex, lint PR.

Lint review completo de un módulo Drupal 11 combinando 4 fuentes en paralelo — PHPStan level 5 + phpstan-drupal, PHPCS Drupal/DrupalPractice, agente drupal-qa (estándares) y agente drupal-security (OWASP). Dos modos — completo (todo el módulo) y diff (solo archivos cambiados vs develop). Genera informe markdown estructurado en la carpeta del IDE con resumen ejecutivo, hallazgos clasificados por severidad, acciones P0/P1/P2 y comandos de verificación. Úsalo siempre que el usuario quiera auditar calidad o seguridad de un módulo Drupal custom, aunque no diga "lint". Triggers — "lint review", "lint del módulo", "auditar módulo Drupal", "revisar módulo custom", "phpstan del módulo", "validar módulo", "qa del módulo", o cuando el usuario pregunta "¿está bien este módulo?", "¿hay errores?", "¿es seguro?". También antes de un release, antes de un PR a develop, o al validar un módulo recién creado.

Persistent development milestone tracker with full context across conversations. Use when: tracking multi-session features, resuming work from a previous conversation, asking 'what's left to do on X' or 'what's pending', breaking work into trackable subtasks, planning complex implementations, updating progress after coding, checking project status, completing a subtask with QA validation, auditing what's missing in the project, syncing milestones with actual code state, or closing/archiving a finished milestone. Also trigger when user references a milestone by name, says 'what did we do last time', 'resume where we left off', 'how far along is X', 'mark this as done', 'milestone done', 'close this milestone', 'what's missing', 'audit the project', 'sync milestones', or wants to plan a feature with subtasks. Commands: /milestone, /milestone <name>, /milestone init, /milestone sync, /milestone start, /milestone done, /milestone update.

Pre-execution assistant that builds a full execution plan with installed skills (✅) AND uninstalled gaps (❌) the task needs, then offers to install missing skills one by one. Use BEFORE starting any multi-step task. Triggers: 'recommend skills', 'what skill should I use', 'advise', 'suggest', 'help me with', or any work instruction involving code, design, planning, debugging, docs, testing, commits, PRs, strategy.

Security auditor for Claude Code skills. Analyzes skills BEFORE installation using a 9-layer threat detection engine (permissions, static patterns, LLM semantic analysis, bundled scripts, data flow, MCP abuse, supply chain, reputation, anti-evasion) with scoring 0-100 and community audit registry. MUST be used whenever the user is about to install a skill — via npx skills add, /find-skills recommendation, /skill-advisor suggestion, or manual request. Also use when user says 'is this skill safe', 'audit this skill', 'check this skill', 'security scan', 'review before installing', or any mention of skill safety/trust/security. Intercept ALL skill installations proactively.

Persistent correction system that captures mistakes from skills or Claude behavior and ensures they never repeat across sessions. Use whenever the user indicates something went wrong — "that's wrong", "fix this", "not what I wanted", "learn from this", "don't do that again", "esto está mal", "aprende de esto", "no hagas eso otra vez", or any correction, complaint, or teaching moment about Claude's output or a skill's behavior. Also use when the user asks to review, list, delete, or manage past corrections, or wants to create an improvement proposal for a skill author. Covers installed skills, general Claude behavior, and cross-session learning persistence.

Track and report local Claude Code usage per request: tokens consumed, estimated cost in €, sessions, projects, and tool breakdown. Use when the user asks about consumption, credits, usage, cost per request, wants to see a report, asks why a specific request was expensive, suspects a process is consuming tokens, wants to optimize their Claude Code usage, or wants to audit tool usage by request. Also triggers on Spanish phrases: 'cuánto me está costando', 'cuántos tokens', 'consumo de hoy', 'qué petición fue cara', 'está consumiendo mucho', 'optimizar consumo', 'reporte de uso', 'ver uso', 'instalar tracker', 'hook no registra'. Commands: /usage-tracker report [hoy|semana|mes|all] [proyecto], /usage-tracker top-requests [hoy|semana], /usage-tracker install, /usage-tracker status