Agent personas

Agent Personas — 142 Expert Personas for Software Development

A comprehensive library of agent personas covering every role in the software development lifecycle. Each persona is a self-contained .md file with YAML frontmatter (name, description, tools, model) defining the agent's identity, capabilities, and constraints.

Architecture

references/
├── agent-organizer.md          # Master orchestrator — analyzes projects, forms agent teams
│
├── beast-mode/                 # Autonomous execution
│   └── beast mode 3.1.md       # Fully autonomous problem-solving, iterates until solved
│
├── core/                       # Core Engineering (13 agents)
│   ├── architect.md            # System design, scalability (opus)
│   ├── planner.md              # Complex feature planning (opus)
│   ├── code-reviewer.md        # Security-aware code review with severity reports
│   ├── code-archaeologist.md   # Deep legacy codebase analysis
│   ├── code-architect.md       # Feature architecture blueprints
│   ├── code-explorer.md        # Traces execution paths, maps dependencies
│   ├── code-simplifier.md      # Simplifies code preserving exact behavior
│   ├── a11y-architect.md       # WCAG 2.2 compliance (web + native)
│   ├── comment-analyzer.md     # Code comment accuracy analysis
│   ├── conversation-analyzer.md# Conversation hook behavior analysis
│   ├── documentation-specialist.md
│   ├── performance-optimizer.md
│   └── type-design-analyzer.md
│
├── development/                # Development Specialists (15 agents)
│   ├── backend-architect.md    # Backend system design
│   ├── frontend-developer.md   # React/frontend across frameworks
│   ├── full-stack-developer.md # End-to-end web apps
│   ├── nextjs-pro.md           # Next.js expert
│   ├── react-pro.md            # React specialist
│   ├── typescript-pro.md       # TypeScript expert
│   ├── python-pro.md           # Python development
│   ├── golang-pro.md           # Go development
│   ├── mobile-developer.md     # React Native + Flutter
│   ├── ui-designer.md          # UI design implementation
│   ├── ux-designer.md          # UX research and design
│   ├── dx-optimizer.md         # Developer experience
│   ├── legacy-modernizer.md    # Legacy code modernization
│   ├── electron-pro.md         # Electron desktop apps
│   └── harmonyos-app-resolver.md # HarmonyOS/ArkTS/ArkUI
│
├── data-ai/                    # Data & AI (11 agents)
│   ├── ai-engineer.md          # LLM apps, RAG, vector search
│   ├── data-engineer.md        # ETL/ELT, data warehouses, streaming
│   ├── data-scientist.md       # SQL, BigQuery, statistics
│   ├── database-optimizer.md   # SQL optimization, indexing, schema
│   ├── graphql-architect.md    # GraphQL schema design
│   ├── ml-engineer.md          # Production ML deployment
│   ├── postgres-pro.md         # PostgreSQL + PgLite
│   ├── prompt-engineer.md      # LLM prompt optimization
│   ├── gan-evaluator.md        # GAN Harness app evaluator (opus)
│   ├── gan-generator.md        # GAN Harness feature generator (opus)
│   └── gan-planner.md          # GAN Harness spec planner (opus)
│
├── infrastructure/             # Infrastructure & DevOps (9 agents)
│   ├── cloud-architect.md      # Multi-cloud (AWS/Azure/GCP)
│   ├── deployment-engineer.md  # CI/CD, Docker/K8s
│   ├── devops-incident-responder.md
│   ├── incident-responder.md   # SRE incident management
│   ├── performance-engineer.md # Infrastructure performance
│   ├── harness-optimizer.md    # Agent harness config optimization
│   ├── homelab-architect.md    # Home/small-lab networks
│   ├── network-architect.md    # Enterprise network design
│   └── network-troubleshooter.md
│
├── orchestrators/              # Project Orchestrators (5 agents)
│   ├── tech-lead-orchestrator.md # Task delegation (opus, max 2 parallel)
│   ├── project-analyst.md      # Tech-stack detection
│   ├── team-configurator.md    # Agent team setup
│   ├── chief-of-staff.md       # Communication triage (opus)
│   └── loop-operator.md        # Autonomous loop monitoring
│
├── quality-testing/            # Quality & Testing (9 agents)
│   ├── code-reviewer.md        # Comprehensive code quality review
│   ├── architect-review.md     # Architecture consistency review
│   ├── debugger.md             # Systematic debugging
│   ├── qa-expert.md            # QA strategy
│   ├── test-automator.md       # Test automation
│   ├── e2e-runner.md           # E2E testing (Playwright/Agent Browser)
│   ├── pr-test-analyzer.md     # PR test coverage review
│   ├── silent-failure-hunter.md # Swallowed error detection
│   └── tdd-guide.md            # TDD enforcement (80%+ coverage)
│
├── spec-agents/                # Spec Pipeline (8 agents)
│   ├── spec-orchestrator.md    # Pipeline coordination
│   ├── spec-analyst.md         # Requirements elicitation
│   ├── spec-architect.md       # Architecture from specs
│   ├── spec-developer.md       # Implementation from specs
│   ├── spec-planner.md         # Task decomposition
│   ├── spec-reviewer.md        # Code quality review
│   ├── spec-tester.md          # Comprehensive test suites
│   └── spec-validator.md       # Final quality validation
│
├── security/                   # Security (2 agents)
│   ├── security-auditor.md     # OWASP, threat modeling, pentest
│   └── security-reviewer.md    # Vulnerability detection & remediation
│
├── reviewers/                  # Language-Specific Reviewers (16 agents)
│   ├── typescript-reviewer.md  ├── python-reviewer.md
│   ├── rust-reviewer.md        ├── go-reviewer.md
│   ├── java-reviewer.md        ├── kotlin-reviewer.md
│   ├── swift-reviewer.md       ├── cpp-reviewer.md
│   ├── csharp-reviewer.md      ├── flutter-reviewer.md
│   ├── fastapi-reviewer.md     ├── database-reviewer.md
│   ├── fsharp-reviewer.md      ├── healthcare-reviewer.md
│   ├── mle-reviewer.md         └── network-config-reviewer.md
│
├── build-resolvers/            # Build Error Resolvers (9 agents)
│   ├── build-error-resolver.md # General build/TS errors
│   ├── cpp-build-resolver.md   ├── dart-build-resolver.md
│   ├── go-build-resolver.md    ├── java-build-resolver.md
│   ├── kotlin-build-resolver.md├── pytorch-build-resolver.md
│   ├── rust-build-resolver.md  └── swift-build-resolver.md
│
├── specialized/                # Framework-Specific Experts (22 agents)
│   ├── django/ (3): django-api-developer, django-backend-expert, django-orm-expert
│   ├── laravel/ (2): laravel-backend-expert, laravel-eloquent-expert
│   ├── python/ (9): python-expert, django-expert, fastapi-expert, ml-data-expert,
│   │              performance-expert, security-expert, testing-expert,
│   │              web-scraping-expert, devops-cicd-expert
│   ├── rails/ (3): rails-backend-expert, rails-api-developer, rails-activerecord-expert
│   ├── react/ (2): react-component-architect, react-nextjs-expert
│   └── vue/ (3): vue-component-architect, vue-nuxt-expert, vue-state-manager
│
├── specialization/             # Domain Specialists (5 agents)
│   ├── api-documenter.md       # OpenAPI specs, Postman
│   ├── documentation-expert.md # Style guides, info architecture
│   ├── doc-updater.md          # Codemaps, doc updates
│   ├── docs-lookup.md          # Context7-based doc lookup
│   └── seo-specialist.md       # Technical SEO audits
│
├── universal/                  # Universal Developers (4 agents)
│   ├── api-architect.md        # API design across stacks
│   ├── backend-developer.md    # Backend across stacks
│   ├── frontend-developer.md   # Frontend across stacks
│   └── tailwind-css-expert.md  # Tailwind CSS v4+
│
├── utility/                    # Utility Agents (2 agents)
│   ├── refactor-agent.md       # Code refactoring specialist
│   └── refactor-cleaner.md     # Dead code cleanup
│
├── opensource/                 # Open Source Pipeline (3 agents)
│   ├── opensource-forker.md    # Fork + strip secrets
│   ├── opensource-sanitizer.md # Verify sanitization
│   └── opensource-packager.md  # OSS packaging
│
├── backend/                    # Backend Architecture (1)
│   └── senior-backend-architect.md
├── frontend/                   # Frontend Architecture (1)
│   └── senior-frontend-architect.md
├── business/                   # Business & Product (1)
│   └── product-manager.md
└── ui-ux/                      # UI/UX Design (1)
    └── ui-ux-master.md

Legacy agents (superseded by newer versions): code-reviewer-local/, security-auditor/, test-engineer/

When to Use

Trigger	Action
Need a specialized agent for a task	Look up the right persona below and load its .md file
Multi-agent team composition	Read agent-organizer.md or tech-lead-orchestrator.md
Code review needed	Load core/code-reviewer.md or language-specific reviewer
Build error	Load matching build-resolvers/ agent
Starting a new project	Use orchestrators/project-analyst.md then team-configurator.md
Debugging	Load quality-testing/debugger.md
Security audit	Load security/security-auditor.md
Architecture decision	Load core/architect.md

How to Use

Step 1: Identify the Right Agent

Match the task to the agent category:

1. Planning & Design → core/architect.md, core/planner.md, core/code-architect.md
2. Implementation → development/ (language/framework specialists)
3. Code Review → core/code-reviewer.md or reviewers/ (language-specific)
4. Testing → quality-testing/tdd-guide.md, quality-testing/e2e-runner.md
5. Debugging → quality-testing/debugger.md
6. Security → security/security-auditor.md
7. Build Errors → build-resolvers/ (language-specific)
8. Infrastructure → infrastructure/ (cloud, deploy, incidents)
9. Data & AI → data-ai/ (ML, databases, prompts)
10. Orchestration → orchestrators/tech-lead-orchestrator.md
11. Open Source → opensource/ (fork, sanitize, package)

Step 2: Load the Agent File

Read the relevant .md file from references/. The YAML frontmatter tells you:

• name — Agent identifier for invocation
• description — What the agent does and when to use it
• tools — Which tools the agent needs access to
• model — Recommended model tier (haiku/sonnet/opus)

Step 3: Compose Multi-Agent Teams

For complex tasks, compose teams using the orchestrators:

Small teams (2-3 agents) — focused tasks:

architect + frontend-developer → Design and implement a UI component

Medium teams (4-6 agents) — feature development:

planner + backend-architect + frontend-developer + code-reviewer + tdd-guide

Large teams (7+ agents) — full project:

tech-lead-orchestrator coordinating:
  ├── architect (system design)
  ├── frontend-developer + ui-designer (frontend)
  ├── backend-architect + database-optimizer (backend)
  ├── security-auditor (security review)
  └── e2e-runner (testing)

Step 4: Orchestration Rules

• Max 2 agents in parallel (per tech-lead-orchestrator rule)
• Personas don't call other personas — the orchestrator or command dispatches
• Model selection: opus for architects/planners, sonnet for developers, haiku for reviewers/documenters
• Independent tasks → parallel execution
• Dependent tasks → sequential execution

Agent Catalog by Lifecycle Phase

Phase	Primary Agents	Support Agents
Define	product-manager, business-analyst	agent-organizer
Plan	architect, planner, tech-lead-orchestrator	project-analyst
Build	frontend-developer, backend-architect, full-stack-developer	specialized/*
Verify	tdd-guide, e2e-runner, test-automator	silent-failure-hunter
Review	code-reviewer, security-auditor, reviewers/*	pr-test-analyzer
Ship	deployment-engineer, cloud-architect	incident-responder
Scale	performance-engineer, performance-optimizer	refactor-cleaner

Rebuttals

Excuse	Rebuttal
"I'll just do it myself"	Specialized agents catch things generalists miss — security vulnerabilities, performance bottlenecks, type design flaws.
"Too many agents, confusing"	Start with one. The agent-organizer.md recommends the right team for you.
"I don't need a reviewer"	Every production bug started as code someone thought was fine. Code review is mandatory.
"Build resolvers are overkill"	They make minimal surgical fixes instead of you thrashing around. Faster and safer.
"Legacy agents are useless"	They're superseded. Use the current versions in core/, quality-testing/, security/.