Syarif Laravel AI Skills

Laravel-focused AI skills for Codex, Claude Code, and any AI coding assistant that can read Markdown files. Install them with npx skills add, use the generated plugin packages, or point a generic assistant at the universal manifest and canonical skills/ folder. The catalog now includes secure memory orchestration, a least-code minimization gate, risk-aware verification, adaptive depth, bounded exploration, and memory discipline so agents can recall useful project context, write the smallest possible diff, match trace and test depth to task risk, and avoid over-exploration or stale memory poisoning.

Skills follow the Agent Skills format.

Language: English | Bahasa Indonesia

What You Get

72 installable Laravel skills - a focused Laravel skill catalog with overlapping topics consolidated into stronger canonical skills.
Least-code minimization gate - least-code enforces YAGNI, reuse, stdlib, native features, installed dependencies, one-liners, and minimum working code before any implementation skill runs.
Risk-aware verification - tasks are classified LOW/MEDIUM/HIGH so trace depth, test scope, and regression surface match the actual risk instead of always running the full suite.
Adaptive depth - small tasks stay small; high-risk tasks get full trace, behavior preservation checks, and failure-path verification without over-exploring.
Bounded exploration - agents stop once the execution path, affected callers, contract, and verification surface are understood, instead of grepping the entire repository.
Memory discipline - memory entries carry lifecycle states, source precedence prevents stale or inferred memory from overriding current code, and checkpoints store only reusable decisions instead of every task result.
Personal core standards - local standards extracted from reviewed Laravel projects and ongoing AI workflows, with client details removed.
Public topic coverage - 54 additional topics mapped from public Laravel skill catalogs without copying third-party skill body text.
Responsive UI testing - a focused Playwright skill for mobile, tablet, desktop, overflow, clipping, tables, modals, navigation, and Livewire state checks.
UI Agent Browser workflow - a frontend/UI/UX skill that combines agent-browser inspection, Playwright checks, target-stack implementation, and backend contract alignment.
Active Secure Memory - memory-management routes conversation, project, user, workflow, and codebase memory through relevance-aware retrieval, local file-backed commands, provenance, anonymization, and secret-safe guardrails.
Simple discovery - npx skills add <repo> --list reads each skills/<folder>/SKILL.md.
Universal assistant support - usable by Codex, Claude Code, Copilot, OpenAI Codex/GPT, Antigravity, Cursor, Windsurf, Zed, Cline, Roo Code, Continue, Kilo Code, Hermes Agent, Aider, OpenCode, Gemini CLI, and generic agents that can read files.

Memory Orchestrator

The memory-management skill is the continuity layer for long-running Laravel work. It helps an agent remember durable decisions, understand related project patterns, continue from a previous checkpoint, and adapt when the current repository proves old memory stale.

It is designed to reduce token waste by retrieving only relevant memory when prior context materially matters: user preferences, project conventions, workflow patterns, conversation checkpoints, and targeted codebase intelligence from a local MCP backend when available. Self-contained syntax, isolated-helper, and generic documentation tasks should skip memory. It also ships an active local backend at skills/memory-management/scripts/memory.mjs, with auto, init, remember, recall, checkpoint, audit, forget, and status commands that default to ~/.ai-memory or AI_MEMORY_ROOT.

Memory entries carry lifecycle states (CURRENT, STALE, SUPERSEDED, TEMPORARY) so stale architecture decisions do not poison later implementation. Source precedence keeps current code authoritative over memory: current code > current config > project docs > explicit project memory > conversation memory > inferred memory. Checkpoints store only reusable durable knowledge—architectural decisions, non-obvious constraints, reusable bug root causes, project conventions, and environment quirks—not every task result. Decision memory records why a choice was made, not just what happened.

Automatic mode is wired into using-laravel-standards as conditional infrastructure, not a specialist slot: before broad exploration, agents decide whether prior project/session/workflow/decision context could materially affect correctness, prefer syarif-memory-management MCP memory_auto, fall back to memory.mjs auto --cwd <project-root> --query "<task intent>" --limit 5 when Node/file access exists, and continue normally when memory is unavailable. memory-management never consumes the 0-2 specialist activation slots. The skill also ships scripts/mcp-server.mjs for stdio MCP tool access and scripts/memory-hook.mjs for lifecycle preflight/checkpoint hooks. For always-on code graph indexing and background watchers, pair this skill with a local codebase memory MCP such as codebase-memory-mcp.

For Hermes-style orchestration, the skill includes references/hermes-orchestrator-profile.md and an orchestrator-profile installer target. This documents portable policy for cross-session memory, external memory providers, skills-on-demand, provider failover, delegated workers, context compression, auxiliary providers, OpenAI-compatible API fronts, messaging gateways, VS Code ACP, Antigravity, and other editor/agent interfaces. The generated profile is non-secret routing metadata; the active host still owns provider switching, subagent execution, compression, and external sync.

Privacy stays first-class: anonymize cross-project knowledge, preserve provenance, verify against source code, and never persist secrets.

Example MCP config:

{
  "mcpServers": {
    "syarif-memory-management": {
      "command": "node",
      "args": ["skills/memory-management/scripts/mcp-server.mjs"],
      "env": {
        "AI_MEMORY_ROOT": "~/.ai-memory"
      }
    }
  }
}

Installer/generator:

If you installed the skills globally with npx skills add ... -g, register Codex memory from the installed skills directory. On Windows PowerShell:

$SkillsRoot = "$env:USERPROFILE\.agents\skills"
$MemoryRoot = "$env:USERPROFILE\.ai-memory"

codex mcp remove syarif-memory-management
codex mcp add syarif-memory-management --env AI_MEMORY_ROOT=$MemoryRoot -- node "$SkillsRoot\memory-management\scripts\mcp-server.mjs"
codex mcp list

The final list should show syarif-memory-management as enabled. If the remove command says the server does not exist, continue with the add command.

If you are working from a local clone of this repository, first cd into the repo root, then use the relative installer/generator commands for detection, previews, or non-Codex targets:

node skills/memory-management/scripts/install-memory-layer.mjs detect
node skills/memory-management/scripts/install-memory-layer.mjs print --target all
node skills/memory-management/scripts/install-memory-layer.mjs install --target ai-agent-tools --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-family --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-copilot-instructions --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-agent-instructions --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-acp-client --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target antigravity --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target antigravity-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target cursor --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target popular-editors --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target windsurf --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target cline --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target roo-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target continue-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target kilo --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target kilo-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target hermes --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target claude-code --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target gemini-cli --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target opencode --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target zed --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target claude --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target orchestrator-profile --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target hooks --apply

Note: npx skills add ... installs the skill files only. It is not active automatic memory integration by itself. To enable the memory MCP/hook layer in a local editor or agent, run the memory installer above explicitly after the skill is installed. Relative node skills/... commands only work when your current directory is the repository root or another directory that contains skills/.

For broad local coverage, vscode-family installs VS Code Stable, VS Code Insiders, VSCodium, and Code - OSS user MCP configs; vscode-acp-client prepares ACP Client for GitHub Copilot, Claude Code, Gemini CLI, and Codex CLI; ai-agent-tools covers Codex CLI, Claude Code, Gemini CLI, and OpenCode; popular-editors covers VS Code, Cursor, Windsurf, and Zed. Use antigravity, kilo, and hermes for their native configs, then orchestrator-profile when you want portable fallback/delegation/compression policy shared by those surfaces.

Installation

Browse the skills.sh page:

https://skills.sh/soden46/syarif-laravel-ai-skills

List available skills:

npx skills add soden46/syarif-laravel-ai-skills --list

Install all skills:

npx skills add soden46/syarif-laravel-ai-skills -s "*" -y

Install all skills globally for Codex:

npx skills add soden46/syarif-laravel-ai-skills -g -a codex -s "*" -y

Interactive install prompt:

Select the Laravel App Skills group to select every skill, then press Enter.

Install one skill:

npx skills add soden46/syarif-laravel-ai-skills -g -a codex -s form-requests -y

Verify installation:

npx skills list --global --agent codex

Updating Skills And Codex Memory

To update installed skills after a new release, rerun the same install command:

npx skills add soden46/syarif-laravel-ai-skills -g -a codex -s "*" -y

memory-management is included when you install with -s "*", but native skill install only updates the skill files. To update the active Codex MCP registration so it points at the latest globally installed memory server, replace the MCP entry:

$SkillsRoot = "$env:USERPROFILE\.agents\skills"
$MemoryRoot = "$env:USERPROFILE\.ai-memory"

codex mcp remove syarif-memory-management
codex mcp add syarif-memory-management --env AI_MEMORY_ROOT=$MemoryRoot -- node "$SkillsRoot\memory-management\scripts\mcp-server.mjs"
codex mcp list

The final codex mcp list output should show syarif-memory-management with Status set to enabled and Args pointing to .agents\skills\memory-management\scripts\mcp-server.mjs.

If codex mcp remove syarif-memory-management says the server does not exist, continue with the codex mcp add ... command.

Codex Plugin

This repository can be submitted directly to codex-marketplace.com because the root .codex-plugin/plugin.json points to canonical skills/.

This repo also ships a Codex plugin marketplace:

codex plugin marketplace add soden46/syarif-laravel-ai-skills --ref main
codex plugin add syarif-laravel-ai-skills@syarif-laravel-ai-skills

For local development from this checkout:

codex plugin marketplace add .
codex plugin add syarif-laravel-ai-skills@syarif-laravel-ai-skills

Claude Code Plugin

The generated Claude plugin package lives at plugins/laravel-app-skills:

claude --plugin-dir ./plugins/laravel-app-skills

Inside Claude Code, reload and call a skill:

/reload-plugins
/laravel-app-skills:using-laravel-standards

For personal-only usage without a plugin package, install skills directly:

npx skills add soden46/syarif-laravel-ai-skills -g -a claude-code -s "*" -y

Universal AI Usage

For assistants without native skill/plugin support, add this repo to the workspace and start with:

Read AGENTS.md, then use skills/using-laravel-standards/SKILL.md as the entry skill for this Laravel repo. Load focused skills from skills/<skill-name>/SKILL.md only when relevant to the task.

Machine-readable metadata lives in agent-skills.json. Full assistant-specific notes are in docs/UNIVERSAL_USAGE.md.

Quick Start

After installation, ask your assistant to start with:

Use using-laravel-standards for this Laravel repo.

For focused tasks, call a smaller skill directly:

Use least-code to make the smallest possible change.
Use form-requests to move this validation out of the controller.
Use database-transactions to review this checkout flow.
Use ui-agent-browser to design and implement this frontend against the Laravel backend contract.
Use memory-management to recall relevant project and workflow memory before implementing this related feature.
Use quality-checks before final handoff.

What's Inside

Core Local Standards

using-laravel-standards - entry point and skill selector.
memory-management - active secure memory for conversation, project, user, workflow, and codebase context.
least-code - minimization gate that enforces YAGNI, reuse, stdlib, native features, installed dependencies, one-liners, change surface budget, behavior preservation, risk classification, and root-cause confidence before implementation.
extract-laravel-standards - audit finished projects and propose reusable standards.
architecture - Laravel-native architecture decisions.
controller-cleanup - thin controllers and route boundaries.
form-requests - validation and request authorization.
actions-and-services - use-case boundaries, services, and integrations.
database-transactions - atomic writes, locks, and after-commit work.
eloquent-patterns - explicit models, relationships, and query shape.
livewire-development - version-aware Livewire components, state, security, performance, and tests.
queues-and-jobs - jobs, schedules, workers, and operational safety.
security - authorization, request forgery, uploads, secrets, and APIs.
testing - focused tests and verification before handoff.

Topic Coverage

Start Here - using-laravel-standards, runner-selection, daily-workflow, memory-management, brainstorming, writing-plans, executing-plans, laravel-specialist, laravel-11-12-app-guidelines.
Architecture - architecture, module-per-menu, actions-and-services, controller-cleanup, routes-best-practices, ports-and-adapters, interfaces-and-di, strategy-pattern, template-method-and-plugins, complexity-guardrails, constants-and-configuration, custom-helpers.
HTTP and Security - form-requests, api-resources-and-pagination, api-surface-evolution, policies-and-authorization, rate-limiting, request-forgery-protection, security.
Data and Eloquent - migrations-and-factories, eloquent-patterns, eloquent-relationships, database-transactions, data-chunking-large-datasets.
Performance - laravel-database-optimization, performance-caching, performance-eager-loading, performance-select-columns.
UI and Admin - blade-components-and-layouts, ui-agent-browser, livewire-development, nova-resource-patterns, internationalization-and-translation.
Ops and Integrations - queues-and-jobs, horizon-metrics-and-dashboards, task-scheduling, filesystem-uploads, http-client-resilience, integrate-whatsapp-baileys-laravel, exception-handling-and-logging, config-env-storage.
Quality - testing, tdd-with-pest, controller-tests, e2e-playwright, responsive-ui-testing, quality-checks, documentation-best-practices, dependencies-trim-packages.
Modern Laravel - ai-sdk, vector-search, php-attributes, upgrade-13.
Prompting - effective-context, prompt-structure, debugging-prompts, code-review-requests, iterating-on-code, specifying-constraints, using-examples-in-prompts, laravel-prompting-patterns, extract-laravel-standards.

Run the list command for the complete catalog, or read docs/SUPERPOWERS_SKILL_MAPPING.md and docs/LARAVEL_SKILLS_CLOUD_MAPPING.md for public-topic mapping.

How It Works

Each installable skill lives at skills/<skill-folder>/SKILL.md.
The name frontmatter uses <skill-folder>.
The description frontmatter appears in npx skills add <repo> --list.
The tags frontmatter includes laravel and php for Laravel Skills import.
skills.sh.json groups the skills shown on the skills.sh repository page.
plugin-groups.json assigns every skill to an installable plugin bundle.
agent-skills.json exposes neutral metadata for generic AI assistants and integration tools.
.codex-plugin/plugin.json exposes the repo root as a codex-marketplace.com artifact and points directly to skills/.
.agents/plugins/marketplace.json exposes the repo as a Codex/ChatGPT plugin marketplace.
plugins/laravel-app-skills/.claude-plugin/plugin.json makes the generated bundle usable with claude --plugin-dir after npm run sync.
npm run sync regenerates agent-skills.json, .codex-plugin/plugin.json, .agents/plugins/marketplace.json, .claude-plugin/marketplace.json, and local generated plugin output.
Generated plugins/<plugin>/skills/ copies are ignored in Git so marketplace submissions stay under the 128-file scan limit.
package.json is only for local helper scripts. Users install from GitHub with npx skills add, not npm install.

Marketplace Indexing

This repository is designed for skills.sh and Laravel Skills discovery:

skills.sh sees GitHub repositories after someone installs from the repo with the skills CLI.
skills.sh repo pages are cached, so updates can take time after a valid install.
Each skill includes laravel and php tags for Laravel Skills classification.
skills.laravel.cloud imports from the skills.sh ecosystem and lists Laravel/PHP skills after its import and security-audit pass.
Run npx skills add soden46/syarif-laravel-ai-skills -s "*" -y after pushing a release to refresh skills.sh telemetry.

Local Development

npm run validate
npm run list
npm run install:local
npm run sync

To sync missing public topics from Superpowers Laravel:

npm run import:superpowers
npm run validate
npx skills add . --list

Use docs/ADDING_SKILLS.md as the standard for every future skill.

For contribution rules, privacy checks, and PR checklist, read CONTRIBUTING.md.

Reference

This README follows the readable documentation pattern from jpcaparas/superpowers-laravel: clear summary, install commands, quick start, catalog overview, workflow notes, and release links.

Related files:

License

MIT License - see LICENSE.

Skill Laravel-focused AI untuk Codex, Claude Code, dan AI coding assistant yang bisa membaca file Markdown. Install dengan npx skills add, pakai package plugin yang di-generate, atau arahkan assistant generic ke universal manifest dan folder skills/. Katalog ini sekarang menyertakan memory orchestration yang aman, gate minimisasi least-code, verifikasi yang sadar risiko, kedalaman adaptif, eksplorasi yang dibatasi, dan disiplin memory sehingga agent bisa mengingat konteks proyek yang berguna, menulis diff terkecil yang benar, menyesuaikan kedalaman trace dan test dengan risiko task, dan menghindari over-exploration atau memory poisoning.

Yang Didapat

72 skill Laravel installable - katalog skill Laravel yang fokus dengan topik tumpang tindih digabung menjadi skill canonical yang lebih kuat.
Gate minimisasi least-code - least-code menerapkan YAGNI, reuse, stdlib, fitur native, dependency terinstal, one-liner, change surface budget, behavior preservation, risk classification, dan root-cause confidence sebelum skill implementasi lain dijalankan.
Verifikasi yang sadar risiko - task diklasifikasikan LOW/MEDIUM/HIGH sehingga kedalaman trace, scope test, dan regression surface sesuai dengan risiko aktual, bukan selalu menjalankan full suite.
Kedalaman adaptif - task kecil tetap kecil; task berbahaya dapat full trace, behavior preservation checks, dan failure-path verification tanpa over-exploration.
Eksplorasi yang dibatasi - agent berhenti ketika execution path, affected callers, contract, dan verification surface sudah dipahami, bukan menggrep seluruh repository.
Disiplin memory - memory entries memiliki lifecycle states, source precedence mencegah stale/inferred memory meng override current code, dan checkpoint menyimpan hanya reusable decisions.
Standar inti pribadi - standar lokal dari proyek Laravel yang sudah direview dan workflow AI yang terus berkembang, tanpa detail client.
Cakupan topik publik - 54 topik tambahan yang dimapping dari katalog skill Laravel publik tanpa menyalin isi body skill pihak ketiga.
Responsive UI testing - skill Playwright khusus untuk mobile, tablet, desktop, overflow, clipping, tabel, modal, navigasi, dan state Livewire.
Workflow UI Agent Browser - skill frontend/UI/UX yang menggabungkan inspeksi agent-browser, pengecekan Playwright, implementasi sesuai stack target, dan alignment kontrak backend.
Active Secure Memory - memory-management mengatur conversation, project, user, workflow, dan codebase memory dengan retrieval relevan, command lokal file-backed, provenance, anonymization, dan guardrail anti-secret.
Discovery sederhana - npx skills add <repo> --list membaca setiap skills/<folder>/SKILL.md.
Support AI universal - bisa dipakai Codex, Claude Code, Copilot, OpenAI Codex/GPT, Antigravity, Cursor, Windsurf, Zed, Cline, Roo Code, Continue, Kilo Code, Hermes Agent, Aider, OpenCode, Gemini CLI, dan agent generik yang bisa membaca file.

Memory Orchestrator

Skill memory-management adalah lapisan kontinuitas untuk kerja Laravel yang panjang. Agent bisa mengingat keputusan tahan lama, memahami pola proyek terkait, melanjutkan dari checkpoint sebelumnya, dan beradaptasi ketika repository saat ini membuktikan memory lama sudah stale.

Skill ini dirancang untuk menghemat token dengan hanya mengambil memory yang relevan ketika prior context memang material: preferensi user, konvensi proyek, pola workflow, checkpoint percakapan, dan codebase intelligence terarah dari backend MCP lokal ketika tersedia. Task syntax self-contained, helper isolated, dan pertanyaan dokumentasi generic harus skip memory. Skill ini juga membawa backend lokal aktif di skills/memory-management/scripts/memory.mjs, dengan command auto, init, remember, recall, checkpoint, audit, forget, dan status yang default ke ~/.ai-memory atau AI_MEMORY_ROOT.

Memory entries memiliki lifecycle states (CURRENT, STALE, SUPERSEDED, TEMPORARY) sehingga architecture decisions lama tidak menyesatkan implementasi berikutnya. Source precedence menjamin current code selalu di atas memory: current code > current config > project docs > explicit project memory > conversation memory > inferred memory. Checkpoint menyimpan hanya reusable durable knowledge—architectural decisions, non-obvious constraints, reusable bug root causes, project conventions, environment quirks—bukan semua hasil task. Decision memory records why a choice was made, bukan cuma what happened.

Mode otomatis sudah disambungkan ke using-laravel-standards sebagai infrastruktur conditional, bukan slot specialist: sebelum eksplorasi luas, agent memutuskan apakah konteks proyek/session/workflow/decision lama bisa mempengaruhi correctness, mengutamakan MCP syarif-memory-management tool memory_auto, fallback ke memory.mjs auto --cwd <project-root> --query "<task intent>" --limit 5 jika Node/file access tersedia, dan lanjut normal jika memory unavailable. memory-management tidak pernah mengonsumsi slot aktivasi specialist 0-2. Skill ini juga membawa scripts/mcp-server.mjs untuk tool MCP stdio dan scripts/memory-hook.mjs untuk lifecycle hook preflight/checkpoint.

Privasi tetap jadi aturan utama: anonymize knowledge lintas proyek, simpan provenance, verifikasi terhadap source code, dan jangan pernah persist secret.

Contoh config MCP:

{
  "mcpServers": {
    "syarif-memory-management": {
      "command": "node",
      "args": ["skills/memory-management/scripts/mcp-server.mjs"],
      "env": {
        "AI_MEMORY_ROOT": "~/.ai-memory"
      }
    }
  }
}

Installer/generator:

Kalau skill di-install global dengan npx skills add ... -g, register memory Codex dari folder skill yang sudah terpasang. Di Windows PowerShell:

$SkillsRoot = "$env:USERPROFILE\.agents\skills"
$MemoryRoot = "$env:USERPROFILE\.ai-memory"

codex mcp remove syarif-memory-management
codex mcp add syarif-memory-management --env AI_MEMORY_ROOT=$MemoryRoot -- node "$SkillsRoot\memory-management\scripts\mcp-server.mjs"
codex mcp list

Output akhir harus menampilkan syarif-memory-management sebagai enabled. Kalau command remove bilang server belum ada, lanjutkan saja ke command add.

Kalau memakai clone lokal repository ini, masuk dulu ke root repo dengan cd, baru pakai command installer/generator relatif untuk deteksi, preview, atau target selain Codex:

node skills/memory-management/scripts/install-memory-layer.mjs detect
node skills/memory-management/scripts/install-memory-layer.mjs print --target all
node skills/memory-management/scripts/install-memory-layer.mjs install --target ai-agent-tools --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-family --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-copilot-instructions --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-agent-instructions --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-acp-client --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target vscode-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target antigravity --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target antigravity-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target cursor --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target popular-editors --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target windsurf --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target cline --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target roo-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target continue-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target kilo --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target kilo-workspace --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target hermes --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target claude-code --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target gemini-cli --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target opencode --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target zed --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target claude --apply
node skills/memory-management/scripts/install-memory-layer.mjs install --target hooks --apply

Catatan: npx skills add ... hanya memasang file skill. Itu belum berarti integrasi memory otomatis sudah aktif. Untuk mengaktifkan memory MCP/hook di editor atau agent lokal, jalankan installer memory di atas secara eksplisit setelah skill terpasang. Command relatif node skills/... hanya jalan kalau current directory adalah root repository atau folder lain yang punya subfolder skills/.

Untuk cakupan lokal yang luas, vscode-family memasang config MCP user untuk VS Code Stable, VS Code Insiders, VSCodium, dan Code - OSS; vscode-acp-client menyiapkan ACP Client untuk GitHub Copilot, Claude Code, Gemini CLI, dan Codex CLI; ai-agent-tools mencakup Codex CLI, Claude Code, Gemini CLI, dan OpenCode; popular-editors mencakup VS Code, Cursor, Windsurf, dan Zed. Pakai antigravity, kilo, dan hermes untuk config native masing-masing.

Instalasi

Lihat halaman skills.sh:

https://skills.sh/soden46/syarif-laravel-ai-skills

Lihat daftar skill:

npx skills add soden46/syarif-laravel-ai-skills --list

Install semua skill:

npx skills add soden46/syarif-laravel-ai-skills -s "*" -y

Install semua skill global untuk Codex:

npx skills add soden46/syarif-laravel-ai-skills -g -a codex -s "*" -y

Prompt install interaktif:

Pilih group Laravel App Skills untuk memilih semua skill, lalu tekan Enter.

Install satu skill:

npx skills add soden46/syarif-laravel-ai-skills -g -a codex -s form-requests -y

Cek hasil install:

npx skills list --global --agent codex

Update Skill Dan Memory Codex

Untuk update skill yang sudah terpasang setelah release baru, jalankan ulang command install yang sama:

npx skills add soden46/syarif-laravel-ai-skills -g -a codex -s "*" -y

memory-management ikut terpasang kalau memakai -s "*", tapi native skill install hanya mengupdate file skill. Supaya registrasi MCP Codex aktif ikut menunjuk ke memory server global terbaru, replace entry MCP-nya:

$SkillsRoot = "$env:USERPROFILE\.agents\skills"
$MemoryRoot = "$env:USERPROFILE\.ai-memory"

codex mcp remove syarif-memory-management
codex mcp add syarif-memory-management --env AI_MEMORY_ROOT=$MemoryRoot -- node "$SkillsRoot\memory-management\scripts\mcp-server.mjs"
codex mcp list

Output akhir codex mcp list harus menampilkan syarif-memory-management dengan Status enabled dan Args mengarah ke .agents\skills\memory-management\scripts\mcp-server.mjs.

Kalau codex mcp remove syarif-memory-management bilang server belum ada, lanjutkan saja ke command codex mcp add ....

Plugin Codex

Repository ini bisa langsung disubmit ke codex-marketplace.com karena root .codex-plugin/plugin.json menunjuk ke canonical skills/.

Repo ini juga menyediakan marketplace plugin Codex:

codex plugin marketplace add soden46/syarif-laravel-ai-skills --ref main
codex plugin add syarif-laravel-ai-skills@syarif-laravel-ai-skills

Untuk development lokal dari checkout ini:

codex plugin marketplace add .
codex plugin add syarif-laravel-ai-skills@syarif-laravel-ai-skills

Plugin Claude Code

Package plugin Claude yang digenerate ada di plugins/laravel-app-skills:

claude --plugin-dir ./plugins/laravel-app-skills

Di dalam Claude Code, reload lalu panggil skill:

/reload-plugins
/laravel-app-skills:using-laravel-standards

Kalau cuma dipakai sendiri tanpa package plugin, install skill langsung:

npx skills add soden46/syarif-laravel-ai-skills -g -a claude-code -s "*" -y

Penggunaan AI Universal

Untuk assistant yang belum punya support skill/plugin native, masukkan repo ini ke workspace lalu mulai dengan:

Read AGENTS.md, then use skills/using-laravel-standards/SKILL.md as the entry skill for this Laravel repo. Load focused skills from skills/<skill-name>/SKILL.md only when relevant to the task.

Metadata machine-readable ada di agent-skills.json. Catatan lengkap per assistant ada di docs/UNIVERSAL_USAGE.md.

Quick Start

Setelah install, minta assistant mulai dari:

Use using-laravel-standards for this Laravel repo.

Untuk task yang lebih spesifik, panggil skill kecil langsung:

Use least-code to make the smallest possible change.
Use form-requests to move this validation out of the controller.
Use database-transactions to review this checkout flow.
Use ui-agent-browser to design and implement this frontend against the Laravel backend contract.
Use memory-management to recall relevant project and workflow memory before implementing this related feature.
Use quality-checks before final handoff.

Isi Repository

Standar Inti Lokal

using-laravel-standards - entry point dan pemilih skill.
memory-management - memory aktif yang aman untuk conversation, project, user, workflow, dan codebase context.
least-code - gate minimisasi yang menerapkan YAGNI, reuse, stdlib, fitur native, dependency terinstal, one-liner, change surface budget, behavior preservation, risk classification, dan root-cause confidence sebelum implementasi.
extract-laravel-standards - audit proyek selesai dan usulkan standar reusable.
architecture - keputusan arsitektur Laravel-native.
controller-cleanup - controller tipis dan batas route.
form-requests - validasi dan request authorization.
actions-and-services - use-case boundary, service, dan integrasi.
database-transactions - atomic write, lock, dan after-commit work.
eloquent-patterns - model, relationship, dan query shape yang eksplisit.
livewire-development - komponen Livewire version-aware, state, security, performa, dan test.
queues-and-jobs - job, schedule, worker, dan operational safety.
security - authorization, request forgery, upload, secret, dan API.
testing - test yang fokus dan verifikasi sebelum handoff.

Cakupan Topik

Mulai Di Sini - using-laravel-standards, runner-selection, daily-workflow, memory-management, brainstorming, writing-plans, executing-plans, laravel-specialist, laravel-11-12-app-guidelines.
Arsitektur - architecture, module-per-menu, actions-and-services, controller-cleanup, routes-best-practices, ports-and-adapters, interfaces-and-di, strategy-pattern, template-method-and-plugins, complexity-guardrails, constants-and-configuration, custom-helpers.
HTTP dan Security - form-requests, api-resources-and-pagination, api-surface-evolution, policies-and-authorization, rate-limiting, request-forgery-protection, security.
Data dan Eloquent - migrations-and-factories, eloquent-patterns, eloquent-relationships, database-transactions, data-chunking-large-datasets.
Performance - laravel-database-optimization, performance-caching, performance-eager-loading, performance-select-columns

Syarif Laravel AI Skills

At a glance

README

Syarif Laravel AI Skills

What You Get

Memory Orchestrator

Installation

Updating Skills And Codex Memory

Codex Plugin

Claude Code Plugin

Universal AI Usage

Quick Start

What's Inside

Core Local Standards

Topic Coverage

How It Works

Marketplace Indexing

Local Development

Reference

License

Yang Didapat

Memory Orchestrator

Instalasi

Update Skill Dan Memory Codex

Plugin Codex

Plugin Claude Code

Penggunaan AI Universal

Quick Start

Isi Repository

Standar Inti Lokal

Cakupan Topik

Comments (0)