Primitives

fai-accessibility-expert

Accessibility specialist — WCAG 2.2 AA/AAA compliance, ARIA 1.2 patterns, screen reader optimization, keyboard navigation, focus management, and inclusive design for AI chatbots and dashboards.

• responsible-ai
• reliability

fai-ai-agents-expert

AI agents expert — ReAct loops, tool orchestration, memory tiers, multi-agent topologies (supervisor/pipeline/debate/swarm), agent determinism, and production guardrails for autonomous AI systems.

• reliability
• security
• responsible-ai

fai-architect

Senior cloud-native solution architect — Azure Well-Architected Framework alignment, AI system design, multi-service integration, cost modeling, trade-off analysis, and production readiness assessment.

• reliability
• security
• cost-optimization
• performance-efficiency
• +2

fai-autogen-expert

Microsoft AutoGen multi-agent framework — ConversableAgent, GroupChat topologies, code execution sandboxing, nested chat orchestration, human-in-the-loop patterns, and AG2 migration.

• reliability
• responsible-ai
• security

fai-browser-agent

Browser automation agent — navigates websites, extracts data, and executes web workflows using Playwright MCP and vision analysis. Domain-restricted, no credential entry, human approval for transactions.

• security
• reliability
• responsible-ai

fai-collective-researcher

Multi-agent researcher — gathers information from knowledge bases, codebase search, documentation analysis, and web research to ground specialist agents with verified facts.

• reliability
• responsible-ai

fai-collective-reviewer

Multi-agent reviewer — security audit, OWASP LLM Top 10, WAF compliance, code quality, AI safety checks, and PR review with severity-classified feedback.

• security
• responsible-ai
• reliability

fai-deterministic-expert

Deterministic AI specialist — makes AI outputs reproducible, grounded, and auditable with temperature control, seed pinning, JSON schema output, RAG grounding, citation enforcement, and multi-layer hallucination defense.

• reliability
• responsible-ai
• security

fai-play-21-reviewer

Agentic RAG reviewer — retrieval autonomy audit, source selection review, iteration limit verification, citation accuracy check, and reflection quality assessment.

• reliability
• responsible-ai

fai-prompt-engineer

Prompt engineering specialist — system message design, few-shot patterns, chain-of-thought, structured output schemas, anti-hallucination techniques, and evaluation-driven prompt optimization.

• reliability
• responsible-ai
• cost-optimization

a11y-waf

Accessibility standards — WCAG 2.2 AA compliance, semantic HTML, ARIA roles/properties, keyboard navigation, focus management, color contrast, screen reader compatibility, and inclusive design patterns.

• responsible-ai
• reliability

agent-safety

AI agent safety standards — fail-closed defaults, least-privilege tool access, human-in-the-loop for destructive actions, prompt injection defense, budget controls, OWASP LLM Top 10 alignment, and governance audit trails for all agent workflows.

• security
• responsible-ai
• reliability

azure-ai-document-intelligence-waf

Azure AI Document Intelligence standards — model selection, layout analysis, table/form extraction, custom models, batch processing, confidence thresholds, SDK usage, and Bicep deployment.

• reliability
• performance-efficiency
• cost-optimization
• security
• +1

azure-ai-language-waf

Azure AI Language standards — text analytics, NER, sentiment analysis, PII detection, summarization, CLU, custom text classification, healthcare NLP, SDK patterns, and batch processing.

• responsible-ai
• reliability
• performance-efficiency
• security
• +1

azure-ai-speech-waf

Azure AI Speech standards — STT (real-time + batch), TTS (neural voices, SSML, custom voice), speech translation, pronunciation assessment, speaker recognition, WebSocket streaming, SDK patterns, and latency optimization.

• performance-efficiency
• reliability
• security
• cost-optimization
• +1

deterministic-ai

Determinism standards — temperature=0+seed, structured output, validation pipeline, grounding checks.

• reliability
• responsible-ai

prompt-engineering

Prompt engineering standards — system message structure, few-shot patterns, output format enforcement.

• reliability
• responsible-ai