opencode-accomplishments

Publish a React component to the user's team library on 21st.dev so teammates can install it with a single command, or install an existing one. Triggers when the user says things like "publish/share/upload this to 21st", "залей в наш регистр", "опубликуй компонент", "share with team", "make this reusable", "install our Button", "use the team button".

Three-layer animation separation doctrine for Three.js + GSAP + Framer Motion. Three.js owns the 3D scene and objects only. GSAP owns the camera and scroll-linked animation only. Framer Motion owns React UI overlays only. Strict useEffect cleanup, no cross-layer DOM writes, refs-only cross-layer communication.

Fan-out / fan-in deep research pipeline. Plans N sub-queries, dispatches parallel web-search workers (each with own context), dedups URLs, resolves conflicts, synthesizes a single cited markdown report. Use ONLY when user explicitly asks for "deep", "exhaustive", "comprehensive", or "parallel" research — 10-100x slower than a single web_search. Triggers: deep research, parallel research, orchestrator, fan-out, exhaustive.

Meta-design intelligence layer — a nested skill that orchestrates all design-oriented skills into a unified optimization tree. Acts as the single entry-point for any design request, routing to the optimal sub-skill based on context, complexity, and output target. Integrates the Supanova premium aesthetic suite from uxjoseph (skills.sh/uxjoseph) alongside existing design, animation, and visual composition skills.

API gateway orchestration skill — design, deploy, and manage API gateways with rate limiting, authentication, request transformation, and observability for AI-powered microservices. Use when use when building api gateways, managing rate limiting for ai services, implementing authentication layers for multi-provider ai endpoints, or designing microservice api architectures Triggers on keywords: api, gateway, rate-limit, authentication, microservice, proxy, observability.

Unified billing and cost management for multi-proxy stacks — aggregates usage, cost, and quota data across 9Router, Owl, Antigravity, and Triune proxies with per-provider spend tracking, budget alerts, and optimization recommendations. Use when use when managing billing across multiple proxy stacks, tracking ai model costs across providers, or optimizing proxy stack spending Triggers on keywords: billing, cost, proxy, quota, budget, optimization, spend.

OpenCode-optimized Owl proxy installer — automated setup of Owl proxy defense stack for OpenCode CLI with pre-configured tier routing, anti-bot escalation, and credential vaulting. Use when use when installing owl proxy for opencode cli, configuring anti-bot defense for ai coding agents, or setting up tier-based routing for opencode workflows Triggers on keywords: opencode, owl, install, proxy, setup, anti-bot, tier-routing.

OpenHuman-optimized Owl proxy installer — automated setup of Owl proxy defense stack for OpenHuman agent platform with human-in-the-loop escalation, consent-gated tier progression, and privacy-first credential management. Use when use when installing owl proxy for openhuman platform, configuring privacy-first anti-bot defense for human-agent workflows, or setting up consent-gated proxy escalation Triggers on keywords: openhuman, owl, install, proxy, privacy, consent, human-in-the-loop.

Go-based open relay proxy server — high-performance, concurrent relay proxy written in Go with connection pooling, load balancing, health checking, and zero-config auto-discovery for AI provider endpoints. Use when use when building high-performance proxy relays, implementing go-based ai model routing, or setting up zero-config relay servers for multi-provider ai access Triggers on keywords: go, relay, proxy, openrelay, high-performance, concurrent, load-balance.

Unified meta-CLI wrapper for OpenCode + Claude Code + Codex + Gemini CLI — one command for four agents with 9Router tier routing and unified credential vault. Use when use when managing multiple ai coding clis, setting up unified routing across opencode/claude/codex/gemini, or building credential vaults Triggers on keywords: cli, unified, opencode, claude, codex, gemini, owl.

AI routing gateway with RTK compression, tier-based provider fallback, circuit breaker, and real-time dashboard — routes requests through free-tier providers first, paid only on fallback. Use when use when setting up ai model routing, managing provider tiers, or implementing circuit breaker patterns for api resilience Triggers on keywords: routing, gateway, 9router, fallback, circuit-breaker, rtk.

Break into steps — task decomposition, work breakdown structures, and sequential execution planning. Ships incrementally. Use when use when decomposing tasks, planning work breakdowns, or shipping incrementally Triggers on keywords: agent, decompose, steps, planning, incremental.

AUDIT animation code against the Design OS standards — bundle budget, accessibility, performance, memory, and framework integration. Use when reviewing animation code, running CI checks, pre-commit validation, or when the user asks "does this animation code look right" or "audit my animations". Also trigger for animation-related PRs, before deployment, or when onboarding new team members to animation standards. This is the FINAL agent in the pipeline — code does not ship without passing this agent.

Build systems — architecture design, infrastructure planning, and system construction. Builds things that last. Use when use when designing architecture, planning infrastructure, or building systems Triggers on keywords: agent, architecture, infrastructure, systems, builder.

DECIDE whether to animate, which tool to use, and what the budget impact is BEFORE writing any animation code. Use when the user mentions animation, motion, transition, or any visual effect. Also trigger BEFORE any animation task — this agent gates all animation work. This is the FIRST agent in the animation pipeline. No animation code may be written without passing through this agent.

Creative solutions — lateral thinking, analogical reasoning, and innovative problem-solving. Finds elegant unconventional paths. Use when use when finding creative, lateral, or unconventional solutions to problems Triggers on keywords: agent, creative, lateral, innovative, elegant.

Big picture — strategic vision, cross-domain patterns, and second-order thinking. Sees the forest AND the trees. Use when use when seeing the big picture, finding cross-domain patterns, or strategic vision Triggers on keywords: agent, strategy, vision, patterns, cross-domain.

Cross-field connections — knowledge synthesis, interdisciplinary bridging, and pattern transfer. Connects dots across domains. Use when use when synthesizing knowledge across domains or bridging interdisciplinary gaps Triggers on keywords: agent, synthesis, interdisciplinary, connections, knowledge.

MONITOR animation performance in production, detect regressions, and maintain animation standards as the codebase evolves. Use when animation code has been deployed, when setting up monitoring, or when investigating production animation issues. Also trigger for animation regression reports, performance degradation over time, or when updating animation libraries. Think like an owl: observe patterns over time, not just snapshots.

ORCHESTRATE the complete animation lifecycle — from decision to maintenance — using 6 specialized sub-agents. Use when ANY animation work is requested. This agent decomposes the task, routes to sub-agents, and coordinates the pipeline. Also trigger for "animate this", "fix animation", "optimize animation", "audit animation", or any animation-related request. This is the TOP-LEVEL agent. All animation work flows through here.

OPTIMIZE animation performance AFTER profiling — apply targeted fixes to the ACTUAL bottleneck, not guessed improvements. Use when the profiler agent has identified a bottleneck, or when the user provides profiling data and asks for optimization. Also trigger AFTER profiling is complete — never before. This agent prevents premature optimization. Think like an owl: verify the fix worked before declaring success.

Deep analysis — systematic decomposition, evidence evaluation, and convergent thinking. Finds what others miss. Use when use when performing deep analysis, evaluating evidence, or convergent thinking Triggers on keywords: agent, analysis, decompose, converge, critical.

PROFILE animation performance BEFORE optimizing — identify the ACTUAL bottleneck using Chrome DevTools, React Profiler, and custom metrics. Use when the user says "animation is slow", "laggy", "janky", "dropping frames", or wants to optimize animation performance. Also trigger BEFORE any optimization work — this agent prevents "premature optimization" by finding the real bottleneck. Think like an owl: slow, observant, analytical. Measure twice, optimize once.

Multiply ideas — rapid ideation, brainstorming, and divergent thinking. Generates many options quickly, then clusters and prioritizes. Use when use when brainstorming, ideating, or generating many creative options quickly Triggers on keywords: agent, ideation, brainstorm, diverge, creative.

Unified multi-agent role system — Builder, Coder, Council, Orchestrator, Plan, Researcher, Reviewer, Scribe. Use when use when coordinating multi-agent workflows with role delegation Triggers on keywords: agents, roles, orchestration, delegation.

SIMULATE animation end-to-end BEFORE generating code — trace every user interaction, state change, and frame to predict behavior and catch edge cases. Use when planning complex animations, before writing animation code, or when the user wants to understand how an animation will behave across all states. Also trigger for multi-step animations, scroll-driven narratives, or any animation with conditional logic. Think like an owl: trace every path, including the error paths.

Audit codebases for animation anti-patterns, performance violations, accessibility issues, and framework integration errors. Use when reviewing animation code, running CI checks, pre-commit validation, or when the user asks "does this animation code look right", "audit my animations", or "check for animation issues". Also trigger for animation code review, performance audit, bundle size check, accessibility compliance review, or when animation-related PRs are submitted. Always use this skill for any animation code review — do not review animation code without it.

Meta-skill for combining Motion v12 + GSAP v3.13+ with strict boundaries — Motion owns UI layer (modals, gestures, layout), GSAP owns content layer (scroll, text, SVG). Never animate same element with both.

Route animation tasks to the correct animation library (Motion, GSAP, AutoAnimate, CSS, Lottie) based on complexity, framework, bundle budget, and capability requirements. Use when the user asks about animation strategy, which animation library to use, how to choose between Motion and GSAP, animation performance optimization, or when planning animation architecture for a project or design system. Also trigger for "should I use Motion or GSAP", "animation decision tree", "what's the best way to animate this", "animation performance budget", or when the user describes an animation need but hasn't specified a tool. Always use this skill first before any animation implementation — it prevents wrong-tool selection and bundle bloat.

Bold aesthetics for AI-native interfaces, conversational UI, trust-forward design. Use when use when building ai-native or conversational interfaces Triggers on keywords: ai, frontend, trust, conversational.

MITM proxy bridge for free AI model access — intercepts IDE traffic to provide Claude/Gemini model access via multiple account rotation with staggered token refresh. Use when use when setting up free ai model access via mitm proxy, managing staggered oauth token refresh, or bridging ide traffic to ai providers Triggers on keywords: mitm, proxy, claude, gemini, free-tier, antigravity.

Detect and prioritize audit issues — performance, accessibility, monitoring. Use when use when auditing websites for performance, accessibility, or monitoring issues Triggers on keywords: audit, performance, accessibility, monitoring.

Structured brainstorming framework with divergent and convergent phases — generates ideas rapidly, then clusters, prioritizes, and refines them into actionable concepts. Use when use when generating creative ideas, running brainstorming sessions, or exploring solution spaces with structured divergence-convergence Triggers on keywords: brainstorming, ideation, divergent, convergent, creative.

Headful browser automation with natural language control. Use when use when automating browser tasks with natural language commands Triggers on keywords: browser, automation, playwright, testing.

Step-by-step reasoning framework for complex problems. Use when use when solving complex, multi-step problems requiring explicit reasoning Triggers on keywords: reasoning, logic, step-by-step, analysis.

Research open-source repositories to understand how something is built. Use when use when researching open-source code to understand implementation patterns Triggers on keywords: research, codebase, open-source, patterns.

Software composition patterns — dependency injection, strategy pattern, decorator chains, middleware pipelines, and plugin architectures for extensible codebases. Use when use when designing composable architectures, building plugin systems, or implementing dependency injection and strategy patterns Triggers on keywords: composition, patterns, di, strategy, middleware, plugins.

Compress long contexts preserving critical information. Use when use when reducing context length while keeping essential information Triggers on keywords: compression, context, optimization, tokens.

Fetch up-to-date, version-specific documentation and code examples into LLM prompts — eliminates stale or hallucinated API references for any library. Always-current docs. Use when use when looking up current api documentation, fighting hallucinated references, or getting version-specific code examples for any library Triggers on keywords: documentation, context, up-to-date, api, reference, version-specific.

Deploy, monitor, update across GitHub Pages, Vercel, Netlify. Use when use when deploying applications or managing infrastructure Triggers on keywords: deploy, vercel, netlify, github-pages.