MediaOS

Search, install, update, and rate AI agent skills from agentskill.sh (100,000+ skills). Use when the user asks to find skills, install extensions or plugins, discover new capabilities, check what skills are available, or says "how do I do X" when a skill might help. Also handles listing installed skills, checking for updates, rating skills, removing skills, and scanning skills for security issues. Triggers on: /learn, find skill, install skill, search skills, what skills, add capability, get plugin, check skill safety.

Review and improve AI agent skills (SKILL.md files) against best practices from the Agent Skills specification and Anthropic's authoring guidelines. Scores skills on 10 quality dimensions, identifies specific issues, and rewrites problem areas. Use when creating, editing, auditing, or improving agent skills. Triggers on: review skill, improve skill, audit SKILL.md, check skill quality, skill best practices, optimize skill description.

AI-native product thinking and modern product design philosophy. Every product should consider AI copilots, semantic search, intelligent automation, conversational workflows, contextual AI assistance, and recommendation systems. Products must feel modern, premium, intelligent, polished, and production-ready. Use when designing new features, reviewing product roadmap, or evaluating whether AI capabilities would enhance a workflow. Covers Azure OpenAI, OpenAI, Anthropic, RAG pipelines, vector DBs, semantic indexing, and AI workflow orchestration.

Autonomous senior-engineer bug fixing workflow. Investigates independently, inspects logs, traces root causes, reproduces issues, fixes thoroughly, and validates correctness — without asking for unnecessary guidance. Use whenever encountering bugs, failing builds, broken UX, CI failures, test failures, regressions, or any unexpected behavior. Operates the way a staff engineer would: hypothesis-driven, evidence-driven, root-cause-driven.

Aggressive consumer-product improvement doctrine. Runs the Analyze → Critique → Reimagine → Stress Test → Improve → Refine cycle on any shipped or in-flight feature. Evaluates UX friction, cognitive load, perceived intelligence, retention mechanics, delight, AI-native opportunities, and five user perspectives (first-time, daily, power, mobile, impatient). Use after every implementation pass, when auditing existing UI, when expanding a feature, when the user shares product feedback, or proactively to elevate "functional but boring" surfaces into premium consumer-grade experiences. Mandates feature-ecosystem thinking, not isolated screens.

Orchestrate continual learning by delegating transcript mining and AGENTS.md updates to agents-memory-updater. Use when the user asks to mine prior chats, maintain AGENTS.md, or run the continual-learning loop.

Agent Performance Optimization Workflow workflow skill. Use this skill when the user needs Systematic improvement of existing agents through performance analysis, prompt engineering, and continuous iteration and the operator should preserve the upstream workflow, copied support files, and provenance before merging or handing off.

Skill Improvement Methodology workflow skill. Use this skill when the user needs Iteratively improve a Claude Code skill using the skill-reviewer agent until it meets quality standards. Use when improving a skill with multiple quality issues, iterating on a new skill until it meets standards, or automated fix-review cycles instead of manual editing and the operator should preserve the upstream workflow, copied support files, and provenance before merging or handing off.

Self-improving agent system that analyzes conversation quality, identifies improvement opportunities, and continuously optimizes response strategies.

Core execution philosophy for autonomous engineering. Defines agent identity as strategic technical partner, autonomous engineering operator, product architect, systems thinker, and execution engine. Sets staff-engineer quality bar. Use for ALL tasks — implementation, debugging, refactoring, reviewing — to reject shallow thinking, temporary fixes, generic CRUD, poor UX, fragile architecture, and beginner-level patterns.

GitNexus code intelligence — impact analysis, change detection, symbol exploration, knowledge graph queries, and safe refactoring. Indexes the repository as nodes/edges/execution-flows. Use before editing any function/class/method (impact analysis), before committing (change detection), when exploring unfamiliar code, when renaming/refactoring symbols, when debugging unfamiliar bugs, or when the user asks "how does X work" / "what depends on Y" / "is it safe to change Z".

Performance engineering, security, and DevOps expectations. Performance — low latency, minimal bundle size, responsive rendering, caching, GPU acceleration, CUDA, mixed precision, quantization. Security — authentication, authorization, RBAC, secrets, encryption, rate limiting, audit logging, compliance. DevOps — CI/CD, observability, monitoring, deployment automation, secrets management, rollback strategies, environment isolation, cost optimization. Use when designing infrastructure, reviewing production code, hardening systems, or evaluating deployment readiness.

Research-first execution methodology. Before implementing any significant feature, researches competitors, benchmarks UX patterns, analyzes architecture strategies, studies relevant products, and validates assumptions. Integrates GitNexus impact analysis before editing symbols. Use before starting any non-trivial feature, when entering a new domain, or when architectural decisions are required.

Autonomous self-improving skill meta-agent. Discovers, devours, learns from, installs, creates, and publishes AI agent skills across the internet. Use when asked to find skills, improve capabilities, fill skill gaps, create new skills, publish skill repos, route tasks to skills, or run the skill-forge pipeline. Also use when the user says skill-forge, devour, discover skills, find skills, create a skill for X, improve my setup, what skills exist for X, or ship a skill.

Generate a CONTEXT.md shared vocabulary for any project. Reduces agent verbosity by 50-75%, improves variable naming, makes conversations more precise. Interviews the user about domain terms, then produces a structured glossary the agent references every session. Use when starting a new project, onboarding to unfamiliar code, the agent is being too verbose, or conversations have too much jargon confusion. Also use when user says "build context", "shared language", "define terms", or "the agent doesn't understand my project vocabulary".

Enforce schema discipline: queries-first design, index-every-filter, safe migrations, verify-at-scale thinking. Prevents the common failure of designing schemas for demos that break in production. Use when creating tables, planning migrations, designing data models, or when user mentions database, schema, migration, indexing, multi-tenant.

Disciplined diagnosis loop for hard bugs and unexpected behavior. Forces reproduce → minimise → hypothesise → instrument → fix → regression-test. Use when something is broken and you don't know why, when a bug seems intermittent, when a fix didn't work, or when user says diagnose, debug, investigate, or "why is this broken".

Enforces disciplined git practices on every commit. Small atomic commits, conventional messages, branch hygiene, and PR readiness checks. Changes how the agent HANDLES git — not just what commands it knows. Auto-triggers on any git operation. Use when committing, branching, merging, or preparing PRs. Also use when git history is messy, commits are too large, or PRs get rejected for poor git hygiene.

Before creating anything, get grilled about it. Interview yourself relentlessly about every decision until shared understanding is reached. Prevents the #1 failure: building the wrong thing. Use when about to create a skill, start a feature, design a system, or plan any non-trivial work. Also use when user says grill me, question my plan, challenge this, stress-test, or validate my thinking.

Compact the current conversation into a handoff document so another agent or future session can continue the work without losing context. Use when ending a session, switching tasks, the context is getting long, or user says handoff, wrap up, save progress, or "continue this later".

Prevents false progress. Before claiming ANY task is done, forces the agent to provide EVIDENCE — run tests, show output, demonstrate the fix works. Eliminates the #1 agent failure: "says done but isn't". Use when completing any task, fixing bugs, implementing features, or whenever about to say "done" or "fixed" or "implemented". Auto-triggers before any completion claim.

After every code change, automatically review your own work before presenting it. Catches bugs, style issues, and missed requirements BEFORE the user sees them. Use when implementing features, fixing bugs, refactoring, or any code modification. Auto-triggers on code changes. Reduces back-and-forth by catching issues first-pass.

One-time setup for skill-forge. Asks user preferences (search sources, creation style, target platforms, GitHub username, skill categories of interest) and configures all other skills to use those preferences. Run once before using any other skill-forge capability. Use when user says setup, configure, or invokes /setup.

Diagnose and fix web performance through a disciplined measure-first loop. Forces evidence-based optimization: measure → identify bottleneck → fix ONE thing → re-measure → repeat. Prevents guessing. Use when pages load slowly, user mentions performance, LCP, INP, CLS, lighthouse, bundle size, or "make it faster".

Force a perspective shift. Stop looking at the code line-by-line and explain the broader context: how this piece fits in the system, what depends on it, what it depends on, and what the original designer was thinking. Use when lost in unfamiliar code, when changes feel risky because you don't see the full picture, or when user says "zoom out", "big picture", "how does this fit", or "explain the architecture around this".

Spec-driven development using GitHub Spec Kit methodology — structured specifications, architecture alignment, implementation tracking, validation pipelines, technical decision records, and systematic documentation. Use before implementing any significant feature, when designing new subsystems, when architectural decisions are required, or when a PRD/TRD/spec must precede code. Maintains architectural clarity, execution traceability, and requirement consistency.

Aggressive subagent delegation strategy. Uses subagents to isolate context, reduce cognitive overload, parallelize research, and specialize execution. Delegates research, debugging, architecture analysis, dependency exploration, performance analysis, UX research, codebase exploration, security reviews, documentation, and verification tasks. Use whenever a task has 2+ independent threads, requires deep research, or would pollute main context with exploration work.

Task management and self-improvement loop using tasks/todo.md (checkable items, milestones, risks, verification, rollout) and tasks/lessons.md (mistake → cause → prevention → future rule after every user correction). Use at the start of any non-trivial work (create/update todo.md), during work (mark progress, add notes), at completion (verification evidence, lessons learned, remaining debt), and after ANY user correction (append to lessons.md).

UI/UX design intelligence for elite frontend work. Covers design systems, color palettes, typography, animation, accessibility, layout, responsive patterns, and stack-specific best practices for React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui. Use when designing, building, reviewing, fixing, improving, optimizing, or refactoring any UI — components, landing pages, dashboards, admin panels, e-commerce, SaaS, portfolios, blogs, or mobile apps. Especially relevant for glassmorphism, minimalism, dark mode, bento grids, microinteractions, and Linear/Vercel/Stripe-tier polish.

Verification before completion and demand for elegance. Never marks work complete without proof — tests, logs, flows, UX validation, architecture integrity. Requires staff-engineer-level approval standard. Demands elegant solutions and redesigns hacky, repetitive, brittle, or tightly-coupled code. Use before claiming any task done, before commits, and when reviewing any non-trivial implementation.

Plan-mode-first execution and workflow orchestration. Automatically enters PLAN mode for non-trivial tasks (3+ implementation steps, architectural decisions, infrastructure planning, scalability concerns, multiple modules, integrations, complex workflows, or UX systems). Generates detailed execution plans, risks, dependencies, verification strategies. Stops and re-plans on failure. Use when starting any non-trivial task, before touching code, or when work behaves unexpectedly.