Mannay Claude Code

Agents Reference

Complete reference for all Mannay agents (domain specialists).

Overview

Agents are domain specialists invoked by skills or directly. They provide expert knowledge in specific areas.

How Agents Are Invoked

  1. By skills automatically - Skills invoke relevant agents based on task
  2. By the Task tool - Task(subagent_type="agent-name")
  3. By mentioning their domain - Orchestrator detects and invokes

Architecture Agents

backend-architect

Expertise: API design, database schema, authentication, reliability

Invoke when:

Standards:


frontend-architect

Expertise: Component architecture, responsive design, Core Web Vitals

Invoke when:

Standards:


api-designer

Expertise: REST, GraphQL, tRPC, gRPC, contract-first design

Invoke when:

Standards:


system-architect

Expertise: Scalability, component boundaries, long-term decisions

Invoke when:

Standards:


requirements-analyst

Expertise: PRD creation, user stories, scope definition

Invoke when:

Standards:


tech-stack-researcher

Expertise: Technology comparisons, library recommendations

Invoke when:

Standards:


Quality Agents

code-reviewer

Expertise: SOLID principles, code quality, production readiness

Invoke when:

Standards:


typescript-pro

Expertise: Advanced types, type safety, strict mode

Invoke when:

Standards:


accessibility-specialist

Expertise: WCAG 2.1 AA, screen readers, keyboard navigation

Invoke when:

Standards:


performance-engineer

Expertise: Bottleneck identification, optimization, profiling

Invoke when:

Standards:


security-engineer

Expertise: Vulnerability assessment, OWASP Top 10, zero-trust

Invoke when:

Standards:


refactoring-expert

Expertise: Technical debt reduction, clean code, DRY

Invoke when:

Standards:


Documentation Agents

technical-writer

Expertise: Clear documentation, audience-tailored content

Invoke when:

Standards:


documentation-engineer

Expertise: Docs-as-infrastructure, automated generation

Invoke when:

Standards:


learning-guide

Expertise: Progressive teaching, concept explanation

Invoke when:

Standards:


deep-research-agent

Expertise: Comprehensive research, analysis

Invoke when:

Standards:


Agent Chaining

Skills automatically chain agents based on task domain:

Task Domain Agents Chained
User authentication security-engineer → backend-architect → api-designer
UI component frontend-architect → accessibility-specialist → typescript-pro
API endpoint api-designer → backend-architect → security-engineer
Database work backend-architect → typescript-pro
Performance issue performance-engineer → frontend-architect or backend-architect