sheshiyer/skill-clusters
GitHubCreative Frontend — Astro · GSAP · Remotion
Hub-and-spoke cluster for animated, content-driven Astro sites and programmatic video. The orchestrator routes between in-browser animation (GSAP ScrollTrigger, Anime.js, web-motion-library) over an Astro substrate and render-time video (Remotion), with a shared core holding the in-browser-vs-video decision matrix, reduced-motion baseline, performance budget, and Astro hydration boundaries.
Tauri — cross-platform desktop & mobile (Rust + web)
Flagship cluster: 40 Tauri v2 specialists behind one router. The orchestrator places a task on the lifecycle×concern map (scaffold, configure, IPC/Rust↔frontend bridge, windows/chrome, sidecars, plugins, build/sign/distribute, debug/test, migrate) and delegates; tauri-core holds the shared security model — capabilities → permissions → scopes, the IPC trust boundary, CSP, and the process/runtime model — plus the per-platform distribution matrix.
React Native — UI & interaction craft
Hub-and-spoke cluster for toolchain-agnostic React Native UI and interaction: styling, navigation, Reanimated animation, gestures/haptics, screen design, and data fetching. The orchestrator routes by craft; react-native-core holds the runtime model — New Architecture (Fabric/TurboModules/JSI/bridgeless), the Reanimated + Gesture Handler stack, platform differences, and performance. Toolchain/build/ship lives in the expo cluster; native SwiftUI in native-ios.
Raycast — extensions, AI tools, Store publishing
Hub-and-spoke cluster for Raycast. The orchestrator separates the four jobs people mean by 'Raycast' — build an extension/command, make it AI-callable (tools the Raycast AI calls), publish to the Store, or design a Raycast-aesthetic app — and routes; raycast-core holds the extension model (command types, package.json manifest, @raycast/api UI/navigation, data hooks, build/publish). Includes two authored spokes (raycast-ai-extensions, raycast-store-publishing) that fill the AI and publishing gaps
Native Ios (from ECC)
Route a native Apple-platform task to the right skill among the iOS specialists — Swift 6.2 concurrency, SwiftUI architecture, actor persistence, protocol-based DI/testing, on-device FoundationModels, Liquid Glass design, and icon assets. USE WHEN a user is building, modernizing, or shipping a native iOS/macOS app in Swift but hasn't named the specific concern.
Ai Agents Meta (from ECC)
Route an AI-agent engineering task to the right skill among 14 meta specialists — planning a multi-session build, decomposing a plan into an agent chain, orchestrating a squad, running an autonomous loop, auditing/debugging the agent stack, optimizing prompts, and controlling token/cost budget. USE WHEN a user is designing, building, operating, or hardening an LLM-agent system but hasn't named the specific concern.
Quality Eval (from ECC)
Route a quality/testing/evaluation task to the right skill among 12 specialists — eval-driven development, head-to-head agent benchmarking, TDD, the verification gate, E2E (web + Windows desktop), AI-regression and browser QA, error-handling design, performance baselines, write-time linting, and production-readiness audits. USE WHEN a user wants to test, verify, evaluate, benchmark, or harden code or an agent but hasn't named the specific gate.
Frontend Web (from ECC)
Route a web-UI task to the right skill among 18 frontend specialists — framework choice (React/Next, Vue/Nuxt, Angular), build tooling (Vite, Turbopack, Bun), motion, accessibility, design direction, and component conversion. USE WHEN building, reviewing, or shipping a web interface but the specific framework or concern hasn't been named yet.
Python Backend (from ECC)
Route a Python backend task to the right specialist — idiomatic Python, pytest/TDD, the Django stack (patterns, TDD, security, Celery, verification), FastAPI services, and the ML lane (PyTorch, recsys pipelines, MLE workflow). USE WHEN building, testing, securing, or shipping a Python web/ML backend but the specific framework or concern hasn't been named yet.
Jvm (from ECC)
Route a JVM task to the right skill among 16 specialists — Kotlin language/coroutines/Exposed/Ktor, Java coding standards, the Spring Boot stack (patterns, TDD, security, verification), the Quarkus stack (patterns, TDD, security, verification), JPA/Hibernate, and Compose Multiplatform UI. USE WHEN a user is building, testing, securing, or shipping a Kotlin or Java service but hasn't named the framework or the specific concern.
Php Laravel (from ECC)
Route a PHP/Laravel task to the right specialist — architecture patterns (controllers → services → actions, Eloquent, queues), test-driven development with Pest/PHPUnit, the pre-PR/pre-deploy verification loop, security hardening (authz, validation, CSRF, uploads), and package discovery. USE WHEN a user is building, testing, hardening, or shipping a Laravel app but hasn't named the specific concern.
Systems Languages (from ECC)
Route a systems-language task to the right skill among 7 specialists — Go (patterns, testing), C++ (Core Guidelines coding standards, GoogleTest/CTest testing), and Perl (modern 5.36+ patterns, Test2 testing, taint/injection security). USE WHEN a user is writing, reviewing, testing, or hardening Go, C++, or Perl code but hasn't named the language axis or the specific concern.
Backend Architecture (from ECC)
Route a backend task to the right skill among 8 server-side specialists — architecture boundaries (hexagonal/ports-and-adapters), REST API design, HTTP connector building, NestJS structure, MCP servers, deployment/CI-CD, and decision records. USE WHEN a user is designing, building, integrating, or shipping a backend service but hasn't named the specific concern.
Databases Data (from ECC)
Route a database/data task to the right skill among the data-layer specialists — PostgreSQL, MySQL/MariaDB, the Prisma ORM, Redis, ClickHouse analytics, cross-engine migrations, and high-throughput ingestion/ETL. USE WHEN a user is designing, querying, optimizing, migrating, or moving data but hasn't named the specific engine or concern.
Devops Infra (from ECC)
Route a DevOps / infrastructure task to the right specialist among 13 — container & compose patterns, self-hosting clusters, post-deploy canary checks, reproducible dev environments, enterprise network-device diagnostics (Cisco IOS, BGP, interfaces, config validation, Netmiko SSH), and homelab networking (setup, VLANs, Pi-hole DNS, WireGuard VPN). USE WHEN a user is provisioning, diagnosing, segmenting, or shipping infrastructure but hasn't named the specific layer or tool.
Security (from ECC)
Route a security task to the right skill among six specialists — defensive code review, offensive/bounty vulnerability hunting, agent-config (.claude) auditing, source-asset & embedded-dependency scanning, a pre-action fact-forcing gate, and destructive-operation safety locks. USE WHEN a user wants to secure, audit, harden, or attack a codebase or an agent setup but hasn't named the specific concern.
Research Knowledge (from ECC)
Route a research or knowledge task to the right skill among 12 specialists — current-web research, neural discovery, multi-source cited synthesis, systematic literature review, scholarly evaluation, biomedical/patent/genomic databases, codebase onboarding, guided code tours, live docs lookup, and persistent project memory. USE WHEN a user wants to research, investigate, review the literature, look something up, or understand a codebase but hasn't named the specific tool.
Business Content (from ECC)
Route a business-content task to the right skill among 10 specialists — voice profiling, long-form articles, multi-platform content systems, SEO, market/competitor research, launch campaigns, investor decks/memos, investor outreach, lead intelligence, and social-graph ranking. USE WHEN a user is writing, marketing, fundraising, or doing outbound for a product or company but hasn't named the specific concern.
Social Media (from ECC)
Route a social-media task to the right skill among 8 specialists — making the asset (image/video/audio via fal.ai, explainer animation via Manim, demo capture, video editing, video understanding), adapting copy per platform, and publishing/scheduling to X and 13 networks. USE WHEN a user wants to create, adapt, schedule, or publish social content but hasn't named the specific tool or platform.
Agentic Ops (from ECC)
Route a real-world operations task to the right skill among 16 agentic-ops specialists — email, messages, GitHub, Jira, Google Workspace, project flow, unified notifications, terminal/CI, knowledge base, customer + finance billing, automation audit, workspace-surface audit, social connections, and dashboards. USE WHEN an agent must operate, triage, or prove work on a live external surface (inbox, repo, tracker, billing, docs) but the user hasn't named the specific skill.
Blockchain Web3 (from ECC)
Route a blockchain/web3 task to the right specialist among six — Solidity AMM security, autonomous trading-agent security, EVM token-decimal correctness, Node Keccak-256 hashing, prediction-market oracle research, and the cross-cutting market/agent risk review. USE WHEN a user is building, auditing, or wiring an EVM contract, a crypto trading agent, or a prediction-market workflow but hasn't named the specific concern.
Electron — desktop apps (Chromium + Node)
Hub-and-spoke cluster for Electron desktop apps. The orchestrator routes by process × concern (main/renderer IPC, the security checklist, packaging/signing/notarization, auto-update); electron-core holds the defining model — the main/renderer/preload split and the context-isolation security boundary (renderer untrusted, main holds Node/OS power).
Cloudflare (from library)
Route a Cloudflare task to the right spoke among the platform specialists — Workers and Pages compute, the wrangler CLI and binding/config model, storage (KV, R2, D1), Durable Objects for stateful coordination, the Agents SDK and Sandbox SDK, transactional email, and deploy/auth debugging. USE WHEN a user is building, configuring, or shipping something on Cloudflare's edge but hasn't named the specific service.
Growth Seo (from library)
Route an SEO/growth task to the right skill among 8 search specialists — intent research, content briefs, site architecture, schema/structured data, programmatic pages at scale, technical audits, AI-search (AEO/GEO) optimization, and Google's official guidance. USE WHEN a user wants to rank, get found, fix SEO, plan content, or show up in AI answers but hasn't named the specific concern.
Growth Content (from library)
Route a growth/marketing task to the right skill among 13 demand-generation specialists — content strategy, long-form writing, conversion copy and editing, cold and lifecycle email, paid ad creative, lead magnets and free tools, marketing psychology, the idea bank, and launch planning. USE WHEN a user wants to plan, write, convert, or amplify marketing content but hasn't named the specific concern.
Git Pr Ops (from library)
Route a GitHub PR/issue task to the right skill among the git-pr-ops specialists — read-only review, finding-resolution + safe push, deterministic squash merge, raw gh CLI queries, autonomous issue-to-PR fixing, and repo status / next-wave planning. USE WHEN a user is reviewing, preparing, merging, or shipping work through GitHub PRs and issues but hasn't named the specific step.
Browser Automation (from library)
Route a browser / UI automation task to the right driver among the cluster's specialists — debug-first headless Chromium, local-webapp Playwright testing, native macOS UI capture and control, and the interactive-TTY harness that supervises them. USE WHEN a user wants to drive, screenshot, scrape, test, or verify a web page or desktop UI but hasn't named which engine or session model to use.
Documents (from library)
Route a document task to the right skill among the documents specialists — office formats (DOCX/PDF/PPTX/XLSX), docs sites, changelogs, OCR extraction, natural-language PDF edits, NotebookLM, and diagram authoring. USE WHEN a user wants to create, edit, extract from, convert, or publish a document but hasn't named the specific tool or format.
Conductor — closed-loop spec-driven orchestration
The conductor wing: runs conducty's closed loop (shape → plan → trace → execute → verify → improve → review → ship) but dispatches each task to the right skill-cluster via the resolver, activating deferred clusters on demand. Integrates PAI (triage + fail-closed gates), spec-kit (specs), and skill-clusters (the capability resolver). Orchestrator + core enumerate; the 19 conducty-* loop skills load on demand. Vendored from robertbarclayy/conducty (MIT).