TudeOrangBiasa/opencode-workflow

shared-hosting-deployment

Shared hosting/cPanel deploy: private artifact repos, local builds, PHP/Laravel symlinks, rollback. Use when deploying to shared hosting without VPS/Docker.

ask-matt

Ask which skill or flow fits your situation. A router over the skills in this repo.

caveman

Ultra-compressed communication. Cuts token usage ~75% while keeping technical accuracy. Levels: lite/full/ultra/wenyan. Auto-triggers on efficiency request. Use when user says wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk…

setup-matt

Proactive onboarding for all agent profiles. Asks about each profile's needs (coding style, DESIGN.md, test framework, CI/CD, issue tracker, domain docs) one at a time before writing configs. Run when starting a new project or when profiles are missing…

skill-author

Meta-skill: create, audit, validate skills. Bundles write-a-skill principles with extraction criteria. Use when user says "create skill", "audit skill", or asks where to add it.

architecture-decision-records

Use when capture architectural decisions as structured ADRs — context, alternatives, rationale, consequences. Use only when a decision between significant alternatives is made or when reading existing ADRs.

deep-research

Use when multi-source deep research using web search MCP — searches, synthesizes findings, delivers cited reports with source attribution. Use only when the user wants thorough research on any topic with evidence and citations, competitive analysis,…

to-tickets

Break plan/spec into tracer-bullet tickets with blocking edges. Publishes to tracker (local file or real tracker). Use when user wants to convert plan into issues.

code-review

Two-axis diff review: Standards (follows coding style?) and Spec (matches issue/PRD?). Runs parallel sub-agents. Use when user wants review of branch, PR, or asks "review since X".

dependency-audit

Scan project dependencies for known vulnerabilities using Trivy. Use when reviewing PRs, before deployment, or when user asks "are our deps safe?"

skill-spector

Scan AI agent skill files for security vulnerabilities before installing. Use when adding third-party skills from npm, GitHub, or marketplace.

verify-evidence

Verification checklist for AC, test evidence, stuck-loop detection, AFK review. Use when lead needs independent evidence for non-trivial changes or ambiguous results.

containers

Container patterns for Docker and Kubernetes. Use when writing Dockerfiles, docker-compose, K8s manifests, or deployment configs.

database-patterns

Database patterns for PostgreSQL, MySQL, Redis, and Prisma. Use when designing schemas, writing queries, caching strategies, or ORM configuration.

api-connector-builder

API design and connector building. Use when designing REST/GraphQL APIs, building API clients, or integrating external services.

modular-monolith

Architecture patterns including modular monolith and hexagonal architecture. Use when designing module boundaries, ports/adapters, or refactoring for testability.

react-patterns

React patterns, testing, and performance optimization. Use when working with React components, hooks, testing setup, or performance tuning.

database-review

Database review checklist for SQL, migrations, schema design, indexes, RLS, transactions, query performance, and data integrity. Use only when a diff touches database structure, database queries, policies, or persistence semantics.

php-review

PHP and Laravel review checklist for diffs touching PHP, Blade, Livewire, Filament, Eloquent, queues, policies, validation, or Composer. Use only when PHP/Laravel-specific correctness, security, framework conventions, or static-analysis review is needed.

eval

Use when evaluate a session for AI mistakes, user frustration, repeated errors, or scope drift. Use when user says eval session, what went wrong, diagnose session, session audit, post-mortem. Creates report in opencode-workflow/.scratch/evals/. Report only —…

grilling

Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, says "grill me", or uses any 'grill' trigger phrases.

context-budget

Use when audit OpenCode context overhead across agents, skills, MCP servers, and repo instructions. Use only for workflow/context bloat decisions, not normal implementation.

memory-dreaming

Use when consolidate, merge, archive, or clean up OpenViking memory entries. Use when user says consolidate, dream, clean up lessons, memory hygiene, merge duplicates, archive stale. Auto-loads on intent — finds duplicates across sessions, merges, archives…

search-first

Use when research-before-coding for external dependency, integration, or custom utility decisions. Searches npm/PyPI, MCP servers, GitHub, web. Decision matrix: adopt → extend → compose → build. Not for every small code edit.

backend-patterns

Use when designing backend services: layered architecture, error handling, logging, API design, database access, and security patterns.

django

Use only when building, securing, or verifying Django applications — project structure, DRF, ORM optimization, service layer, Celery async tasks, security hardening, pytest TDD, and pre-deployment verification.

fastapi-patterns

Use only when building FastAPI backends — project structure, Pydantic v2 schemas, dependency injection, async handlers, authentication, authorization, transactional service layers, and testing with httpx and pytest.

laravel

Use when building, securing, or deploying Laravel applications — architecture patterns (controllers → services → actions), Eloquent ORM, form requests, queues, events, caching, auth, CSRF/XSS/SQLi prevention, PHPStan, Pint, and pre-deployment verification.

mcp-server-patterns

Use when build MCP servers with Node/TypeScript SDK — tools, resources, prompts, Zod validation, stdio vs Streamable HTTP transport. Use only when implementing or debugging MCP servers.

accessibility

Use when frontend accessibility patterns — semantic HTML, ARIA, form labeling, keyboard navigation, focus management, and screen reader support. Use only when building or reviewing interactive UI components and forms.

angular-developer

Use only when working in Angular projects — components, signals, forms, DI, routing, SSR, accessibility, tests, and CLI tooling.

nextjs-turbopack

Use only when building Next.js apps with Turbopack — configuration, incremental compilation, HMR, Edge Runtime compatibility, and migration from Webpack.

nuxt4-patterns

Use when nuxt 4 patterns for hydration safety, route rules, data fetching, and lazy loading. Use only when debugging hydration mismatches, configuring route rules, or writing SSR-safe data fetching in Nuxt 4 apps.

ui-to-vue

Use only when converting UI mockups, screenshots, or designs into Vue components — composition API, script setup, reactive state, and Tailwind/component libraries.

vite-patterns

Use only when configuring Vite builds: plugin setup, environment handling, SSR, code splitting, proxy configuration, and production optimization.

database-migrations

Use when safe, reversible database migration patterns — schema changes, data backfills, zero-downtime expand-contract, ORM-specific workflows. Use only when creating or reviewing database migrations.

flox-environments

Use only when managing reproducible dev environments with Flox — declarative .flox/manifest.toml, package installation, language runtimes, services, and cross-platform sharing.

codebase-onboarding

Use when analyze an unfamiliar codebase and generate a structured onboarding guide — architecture map, key entry points, conventions, starter AGENTS.md notes. Use only when first opening a project or joining a new repo.

deployment-patterns

Use when deployment workflows — strategies (rolling, blue-green, canary), Docker multi-stage builds, CI/CD pipelines, health checks, rollback, and production readiness checklist. Use only when setting up CI/CD, containerizing apps, or planning production…

git-workflow

Use when git branching strategies, commit conventions, merge vs rebase, PR workflow, conflict resolution, and release management. Use only for branch/commit/PR/release strategy decisions or conflict handling, not routine git status/diff.