Primitives

fai-adr-writer

ADR writer — documents architecture decisions with MADR 3.0 template, context, alternatives, trade-off matrices, consequences, and WAF pillar impact analysis.

• operational-excellence

fai-markdown-expert

Markdown specialist — CommonMark/GFM, agent markup (.agent.md/.instructions.md/SKILL.md), documentation standards (README/ADR/changelog), accessibility, and content structure.

• operational-excellence

fai-play-lifecycle

FAI play lifecycle manager — handles play initialization (scaffold file structure), Bicep deployment, evaluation quality gates, and config tuning for any of the 101 solution plays.

• operational-excellence
• reliability
• cost-optimization

fai-tdd-green

TDD Green phase specialist — writes the minimal implementation to make failing tests pass, no more and no less. Follows the Red-Green-Refactor cycle.

• reliability

fai-tdd-red

TDD Red phase specialist — writes failing tests from requirements before any implementation. Covers happy path, error cases, edge cases, and AI-specific test scenarios.

• reliability

fai-tdd-refactor

TDD Refactor phase specialist — improves code quality while keeping ALL tests green. Applies extract method, reduce complexity, improve naming, and design pattern introduction.

• reliability
• performance-efficiency

fai-cost-tracker

Triggers on Stop (1 script)

fai-governance-audit

Triggers on UserPromptSubmit, SessionStart, Stop (3 scripts)

fai-license-checker

Triggers on Stop (1 script)

fai-output-validator

Triggers on Stop (1 script)

fai-pii-redactor

Triggers on Stop (1 script)

fai-secrets-scanner

Triggers on Stop (1 script)

fai-session-logger

Triggers on SessionStart, Stop, UserPromptSubmit (3 scripts)

fai-token-budget-enforcer

Triggers on SessionStart (1 script)

fai-tool-guardian

Triggers on PreToolUse (1 script)

fai-tool-observer

Triggers on PostToolUse (1 script)

fai-waf-compliance

Triggers on Stop (1 script)

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-coding-patterns

Agent coding standards — ReAct loop, tool selection, memory management, circuit breaker on tools.

• reliability
• security

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

agentic-os-customization

Agentic OS standards — .github folder structure, 7 primitives, 4 layers, plugin packaging.

• operational-excellence

agentic-rag-patterns

Play 21 patterns — Agentic RAG patterns — retrieval decision loop, multi-source routing, query refinement, self-evaluation.

• reliability
• security

ai-glossary

AI glossary consistency — use standard FAI terminology, avoid ambiguous AI terms in code comments.

• operational-excellence

ai-landing-zone-advanced-patterns

Play 11 patterns — Advanced landing zone — multi-region, ExpressRoute, Azure Firewall, Policy as Code, Defender for Cloud.

• reliability
• security