LLM & Agents

DevOps automation for a Next.js (Vercel) frontend, Fly.io backend, and Supabase Postgres. The agent should prefer `just` recipes as the invocation surface. Internally, recipes use a hybrid MCP+CLI approach: Supabase CLI for writes/migrations (local & prod), Fly MCP preferred for ops with flyctl fallback, and Vercel CLI for deploy with optional MCP for env/status. Secrets are loaded exclusively via 1Password CLI; values are never printed.

Automatically generate test scaffolding when user writes new code without tests or mentions needing tests. Supports unit, integration, e2e, and data tests for PHP and JavaScript. Invoke when user mentions "tests", "testing", "coverage", "write tests", or shows new untested code.

Row Level Security policy templates for Supabase - multi-tenant patterns, user isolation, role-based access, and secure-by-default configurations. Use when securing Supabase tables, implementing RLS policies, building multi-tenant AI apps, protecting user data, creating chat/RAG systems, or when user mentions row level security, RLS, Supabase security, tenant isolation, or data access policies.

Create, modify, and manage Claude Code subagents. Use when users request to create new specialized agents, update existing agents, or need templates for common agent patterns. Also use when agents or other skills need to automatically generate subagents for specific domains.

Invoke Claude Code CLI from Python orchestrators and shell scripts. Use when asked to"spawn claude as subprocess", "automate claude cli", "run claude headless","configure --allowedTools", "set up claude hooks", or "parallel claude invocation".Covers permissions, directory access (--add-dir), hooks, sandbox mode, and async patterns.

Common patterns and best practices for multi-agent coordination

Use when instruction drift is likely: multi-step tasks, mid-task detours (tooling upgrades), or when agents must consult MCP memory / Skills / sessions and re-anchor repeatedly. Triggers: 'stay on track', 'follow instructions', 'resume main task', 'detour', 'intermediate task done'.

Integration guide for using detective skills across plugins. Maps agent roles from frontend, bun, and other plugins to appropriate detective skills. Developer agents should use developer-detective, architect agents should use architect-detective, etc.

Expert on Shelby Protocol decentralized blob storage on Aptos blockchain. Coordinates 7 specialized sub-skills covering protocol architecture, SDK usage, smart contracts, CLI tools, RPC infrastructure, dApp building, and storage integration. Triggers on keywords Shelby Protocol, Shelby storage, decentralized storage, Aptos storage, blob storage, Shelby.

Generate a minimalist static image from a prompt, then animate it into a short video. Use when users want a concept image with a simple, obvious motion.

PRD creation with EA agent using AskUserQuestion for clarification, validation with reviewer, then BA extracts stories. NO HITL in this phase - continues to design-phase.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Audits skills for discoverability and triggering effectiveness. Use when analyzing skill descriptions, checking trigger phrase coverage, validating progressive disclosure, reviewing SKILL.md structure, ensuring skill discoverability, testing skill triggering, improving skill descriptions, or evaluating whether a skill will be invoked appropriately. Also triggers when user asks about skill best practices, wants to improve skill discoverability, or needs help with skill structure.

Documentation lifecycle management skill. Activates when user mentions documentation, docs, sync, quality, validation, releases, or setup. Routes to appropriate agent (doc-expert for orchestration, doc-writer for content) and provides gentle reminders about documentation health.

Comprehensive TAG system validator and policy enforcer that monitors, validates, and corrects TAG usage across code, tests, and documentation. Use when ensuring TAG compliance, validating TAG policy violations, analyzing TAG coverage, or when maintaining TAG system integrity and governance.

Exporting fine-tuned models to GGUF format for deployment in llama.cpp, Ollama, and local serving tools. Triggers: gguf, quantization export, llama.cpp, ollama, save_pretrained_gguf, modelfile.

Generate a Sora video from a text prompt via an Azure OpenAI endpoint, then download the resulting .mp4 locally. Use when the user asks to generate a Sora video/video.mp4 from a prompt or wants the generated video saved to disk.

Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking. Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks.

LangChain implementation patterns with templates, scripts, and examples for RAG pipelines

Manage your installed Claude Code skills - install, update, rename, uninstall, and list skills from GitHub URLs. Use when the user wants to install a skill, update a skill, list installed skills, rename a skill, remove/delete/uninstall a skill, or provides a GitHub URL to a skills directory.