Dev-AID

Local LLM integration with llama.cpp, Ollama, and GGUF models for offline inference and self-hosted AI. Use when deploying local models, configuring inference servers, or building offline AI pipelines. Do NOT use for cloud AI APIs (use cloud-api-integration).

Model Context Protocol server implementation for extending Claude with custom tools, resources, and prompts. Use when building MCP servers, MCP tools, or resource providers. Do NOT use for general API design (use api-expert).

Product Requirements Document generator with three modes: interactive builder, validator, and reverse-engineer from existing code. Use when creating PRDs for new projects, validating existing PRDs for completeness, or generating PRDs from existing codebases. Do NOT use for technical architecture docs (use senior-architect) or implementation plans (use plan-review-expert).

WCAG 2.2 compliance patterns for accessible web applications with ARIA, keyboard navigation, and screen reader support. Use when implementing accessibility features, auditing WCAG compliance, or fixing a11y issues. Do NOT use for general UI design decisions (use ui-ux-expert).

API architecture and design patterns spanning REST, GraphQL, and gRPC with versioning strategies, rate limiting, gateway configuration, and security best practices. Use when making architectural decisions about API style, designing API gateways, planning API versioning, or establishing cross-cutting API concerns. Do NOT use for specific framework implementation details like FastAPI routes or Apollo resolvers.

macOS automation with AppleScript and JXA for system scripting, app control, and workflow automation. Use when automating macOS apps, system events, or JXA scripting. Do NOT use for cross-platform scripting (use bash-expert).

Application security expertise covering OWASP Top 10, threat modeling (STRIDE), secure SDLC, and penetration testing patterns. Use when reviewing application code for vulnerabilities, implementing secure coding patterns, or conducting threat analysis. Do NOT use for CI/CD pipeline security (use devsecops-expert).

GitOps with ArgoCD, Argo Workflows, and Argo Rollouts for Kubernetes continuous delivery and progressive deployments. Use when configuring ArgoCD Applications, writing Argo Workflow templates, setting up progressive rollouts, or debugging GitOps sync issues. Do NOT use for general CI/CD pipeline design without ArgoCD (use cicd-expert).

Cross-language async and concurrency patterns for Python asyncio/TaskGroup, TypeScript promises/async-await, and Rust Tokio with proper error handling and cancellation. Use when designing concurrent architectures, debugging race conditions, implementing task pools, or choosing between async patterns across languages. Do NOT use for synchronous-only code without concurrency needs or simple sequential scripts.

Auto-update implementation with cryptographic signature verification, staged rollouts, delta updates, and rollback capabilities for desktop apps. Use when implementing Tauri updater, designing update server infrastructure, configuring update signing keys, or building staged rollout strategies. Do NOT use for manual update flows, app store distribution, or server-side deployment strategies.

Bash scripting with shellcheck compliance, proper quoting, error handling, and shell automation. Use when writing bash scripts, shell commands, or CLI tools. Do NOT use for Python scripts (use python) or macOS automation (use applescript).

Browser automation with Puppeteer, Playwright, CDP, and WebDriver for testing, scraping, and web interaction. Use when automating browsers, web scraping, or end-to-end testing. Do NOT use for native app automation.

Distributed task queues with Celery including Redis/RabbitMQ backends, Flower monitoring, beat scheduling, and task chaining. Use when configuring Celery workers, designing task workflows with chains and chords, setting up periodic tasks, or debugging task failures. Do NOT use for simple Python threading, non-Python task queues, or raw RabbitMQ without Celery (use rabbitmq-expert).

Desktop application CI/CD pipelines with code signing, notarization, artifact management, and release automation. Use when building Tauri/Electron release pipelines, configuring code signing certificates, setting up Apple notarization, or automating desktop app distribution. Do NOT use for server-side CI/CD or web deployment pipelines (use cicd-expert).

CI/CD pipeline design for GitHub Actions and GitLab CI with security gates, caching, artifact management, and deployment strategies. Use when creating workflows, optimizing build pipelines, adding security scanning steps, or configuring deployment environments. Do NOT use for ArgoCD GitOps delivery (use argo-expert) or desktop app code signing pipelines (use ci-cd).

Kubernetes networking with Cilium eBPF including network policies, service mesh, L7 policies, and Hubble observability. Use when writing CiliumNetworkPolicy resources, configuring eBPF-based networking, setting up Hubble monitoring, or implementing zero-trust network segmentation. Do NOT use for non-Cilium CNI plugins like Calico, Flannel, or Weave.

Cloud AI API integration for Claude, GPT, and Gemini with prompt injection prevention, rate limiting, and cost management. Use when integrating cloud LLM APIs, managing API keys, or building AI-powered features. Do NOT use for local LLMs (use llm-integration).

Cross-platform build systems for Windows, macOS, and Linux desktop applications with code signing, cross-compilation, and distribution packaging. Use when configuring multi-platform CI builds, setting up platform-specific code signing, creating installers (MSI, DMG, AppImage), or managing cross-compile toolchains. Do NOT use for web-only builds or server-side deployments.

Database schema design with normalization, indexing strategies, full-text search, and migration patterns. Use when designing schemas, optimizing indexes, planning migrations, or choosing normalization levels. Do NOT use for specific database engines (use sqlite, sqlcipher, or surrealdb-expert).

D-Bus IPC on Linux for system service integration, signal handling, and inter-process communication. Use when implementing D-Bus services, system bus interfaces, or Linux IPC. Do NOT use for macOS or Windows IPC mechanisms.

Deep research workflows with Gemini, Tavily, and Perplexity for comprehensive technical investigation and synthesis. Use when conducting deep technical research, multi-source analysis, or comprehensive investigations. Do NOT use for web scraping (use browser-automation).

Design system architecture with token-based theming, component APIs, and cross-platform consistency. Use when building design systems, defining design tokens, or creating component libraries. Do NOT use for individual component styling (use tailwindcss).

DevSecOps practices with shift-left security automation, SAST/DAST pipeline integration, supply chain security, and compliance as code. Use when embedding security scanning in CI/CD, implementing SBOM generation, configuring dependency vulnerability checks, or designing security gate policies. Do NOT use for application-level security patterns like OWASP Top 10 remediation (use appsec-expert).

Cryptography patterns for encryption, key management, secure key derivation, TLS, and certificate handling. Use when implementing encryption, managing keys, configuring TLS certificates, or choosing cryptographic algorithms. Do NOT use for general authentication patterns (use appsec-expert or kanidm-expert).

Production FastAPI applications with OAuth2/JWT authentication, Pydantic v2 validation, SQLAlchemy 2.0 async patterns, and middleware configuration. Use when building authenticated APIs, implementing database models with SQLAlchemy 2.0, configuring OAuth2 flows, or optimizing async endpoint performance. Do NOT use for Django, Flask, or GraphQL API development (use graphql-expert).

FastAPI development patterns for JARVIS project including async API endpoints, WebSocket communication, and Pydantic model validation. Use when building JARVIS FastAPI services, adding WebSocket endpoints, or writing Pydantic schemas for the JARVIS assistant. Do NOT use for production auth patterns like OAuth2/JWT or SQLAlchemy integration (use fastapi-expert).

GLSL shader programming for fragment and vertex shaders with WebGL integration and visual effects. Use when writing GLSL shaders, shader math, or GPU rendering logic. Do NOT use for high-level 3D frameworks (use threejs-tresjs).

Graph database design with traversal queries, relationship modeling, and Cypher/Gremlin optimization. Use when designing graph schemas, writing traversal queries, or modeling complex relationships. Do NOT use for document or relational databases (use database-design).

GraphQL API design with schema patterns, resolver implementation, real-time subscriptions, and Apollo federation for distributed graphs. Use when designing GraphQL schemas, implementing resolvers, configuring Apollo Gateway/Federation, or adding query depth/complexity limits. Do NOT use for REST API design (use rest-api-design) or gRPC services.

GSAP animation patterns for timelines, tweens, ScrollTrigger, and complex motion sequences. Use when implementing GSAP animations, building scroll-driven effects, or orchestrating timeline sequences. Do NOT use for CSS transitions or non-GSAP animation libraries.

Harbor container registry management with vulnerability scanning, Trivy integration, Notary artifact signing, and RBAC configuration. Use when deploying Harbor, configuring image scanning policies, setting up replication rules, or managing registry access control. Do NOT use for Docker Hub, ECR, GCR, or other non-Harbor registries.

Modern JavaScript with ES6+ patterns, async/await, module systems, and Node.js best practices. Use when writing vanilla JS, working with promises, or building Node.js services. Do NOT use for TypeScript-specific features like generics or branded types (use typescript-expert).

JSON-RPC 2.0 protocol implementation with method routing, batch requests, error codes, and transport-agnostic design. Use when implementing JSON-RPC servers or clients, designing method registries, handling batch requests, or building MCP-compatible transports. Do NOT use for REST APIs (use rest-api-design), GraphQL (use graphql-expert), or gRPC services.

Kanidm identity management server with OAuth2/OIDC provider configuration, LDAP compatibility, and WebAuthn passwordless authentication. Use when deploying Kanidm, configuring OAuth2 clients in Kanidm, setting up LDAP gateway, or implementing WebAuthn with Kanidm as IdP. Do NOT use for general OAuth2/OIDC flows without Kanidm or other identity providers like Keycloak or Auth0.

Linux accessibility automation with AT-SPI2 for GTK/Qt application testing, D-Bus a11y, and UI control. Use when automating Linux desktop apps via AT-SPI or accessibility APIs. Do NOT use for macOS automation (use macos-accessibility).

macOS accessibility automation with AXUIElement for UI testing, element inspection, and system control. Use when automating macOS UI via accessibility APIs or AXUIElement. Do NOT use for Linux accessibility (use linux-at-spi2).

Model quantization techniques for 4-bit/8-bit inference with GGUF format conversion and optimization. Use when quantizing models, converting to GGUF, or optimizing inference size. Do NOT use for model training or fine-tuning.

Motion design principles for UI animations, transitions, easing curves, and micro-interactions. Use when planning animation strategies, choosing easing curves, or designing transition choreography. Do NOT use for GSAP-specific implementation (use gsap).

Nuxt 4 full-stack patterns with server routes, useFetch, hybrid rendering, and runtime config security. Use when building Nuxt applications, configuring SSR/SSG, or implementing server routes. Do NOT use for plain Vue without SSR or routing (use vue3).

OS keychain integration for secure credential storage on macOS Keychain, Windows Credential Manager, and Linux Secret Service. Use when storing secrets in platform keychains, implementing credential management, or securing sensitive tokens. Do NOT use for application-level secrets via environment variables.