Agent Design Philosophy

The Agency agents aren’t generic prompt templates. They’re specialized experts with personality, processes, and proven patterns.

Core Design Principles

Each agent in The Agency is built on five foundational principles:

Strong Personality

Not “I am a helpful assistant” - real character and voice

Clear Deliverables

Concrete outputs, not vague guidance

Success Metrics

Measurable outcomes and quality standards

Proven Workflows

Step-by-step processes that work

Learning Memory

Pattern recognition and continuous improvement

1. Strong Personality

Each agent has a distinct voice and character that shapes their approach.

Why Personality Matters

Generic Approach
Agency Approach

"I am a helpful assistant that can help you with frontend development.
I will do my best to assist you with your coding tasks."

Problem: No distinctive approach, generic responses, no memorable characteristics.

"I'm Frontend Developer - I default to finding performance bottlenecks
and accessibility issues. I don't just implement designs, I optimize
for Core Web Vitals from the start. Every component I build includes
TypeScript types, accessibility attributes, and performance
considerations."

Benefit: Clear expectations, consistent approach, memorable expertise.

Personality Examples

Evidence Collector

Personality: Skeptical, thorough, evidence-obsessedQuote: “I don’t just test your code - I default to finding 3-5 issues and require visual proof for everything. No ‘looks good’ without screenshots.”Approach: Always finds issues, requires comprehensive documentation, maintains high quality standards.

Reddit Community Builder

Personality: Community-focused, authentic, patientQuote: “You’re not marketing on Reddit - you’re becoming a valued community member who happens to represent a brand.”Approach: Value-first engagement, long-term relationship building, genuine participation.

Whimsy Injector

Personality: Playful, creative, strategic, joy-focusedQuote: “Every playful element must serve a functional or emotional purpose. Design delight that enhances rather than distracts.”Approach: Purposeful personality, measurable delight, inclusive design.

Reality Checker

Personality: Skeptical, honest, fantasy-immuneQuote: “Default to ‘NEEDS WORK’ status unless proven otherwise. First implementations typically need 2-3 revision cycles.”Approach: Evidence-based certification, realistic assessments, no fantasy approvals.

2. Clear Deliverables

Every agent produces concrete, measurable outputs.

Deliverable Structure

Code Examples

Real, runnable code with modern best practices:

// Frontend Developer deliverable example
import React, { memo, useCallback } from 'react';
import { useVirtualizer } from '@tanstack/react-virtual';

interface DataTableProps {
  data: Array<Record<string, any>>;
  columns: Column[];
}

export const DataTable = memo<DataTableProps>(({ data, columns }) => {
  // Implementation with performance optimization
  // Accessibility attributes included
  // TypeScript types enforced
});

Templates & Frameworks

Ready-to-use templates for common deliverables:

# Brand Personality Framework
## Personality Spectrum
- Professional Context: [Guidelines]
- Casual Context: [Guidelines]
- Error Context: [Guidelines]
- Success Context: [Guidelines]

Processes & Systems

Step-by-step workflows with decision points:

## Evidence Collection Process
Capture full-page screenshots (desktop/tablet/mobile)
Document interactive states (before/after)
Test user journeys (step-by-step visual proof)
Generate test results report

Documentation

Comprehensive documentation of decisions and outcomes:

# Architecture Decision Record
- Decision: [What was decided]
- Context: [Why this matters]
- Alternatives: [What else was considered]
- Consequences: [Trade-offs and implications]

What Makes a Great Deliverable

Concrete

Not “I’ll help with design” but “Here’s a complete design system with 20 components”

Runnable

Code that actually works, not pseudo-code or concepts

Documented

Explains the ‘why’ behind decisions, not just the ‘what’

Testable

Clear criteria for success and verification

3. Success Metrics

Every agent knows how to measure their effectiveness.

Metric Categories

Quantitative Metrics
Qualitative Metrics
Process Metrics

Numbers-based measurements:Frontend Developer:

Page load times < 3 seconds on 3G
Lighthouse scores > 90
Component reusability > 80%
Zero console errors in production

Growth Hacker:

User acquisition cost < $50
Viral coefficient > 1.2
Conversion rate improvement > 25%
Month-over-month growth > 20%

Evidence Collector:

3-5 issues found per feature
100% screenshot coverage
Device testing across 3+ viewports
Issue fix verification rate > 95%

Using Metrics Effectively

Don’t just track metrics - use them to improve. Each agent uses their metrics to:

Validate approaches - Did the workflow achieve the expected outcomes?
Identify improvements - Where can the process be optimized?
Communicate value - What impact did the agent’s work have?
Set expectations - What should stakeholders expect from this work?

4. Proven Workflows

Each agent follows a battle-tested process.

Workflow Structure

Typical agent workflows follow this pattern:

Example: Frontend Developer Workflow

Project Setup & Architecture

Set up development environment with proper tooling
Configure build optimization and performance monitoring
Establish testing framework and CI/CD integration
Create component architecture and design system foundation

Deliverable: Project scaffold with architecture docs

Component Development

Create reusable component library with TypeScript types
Implement responsive design with mobile-first approach
Build accessibility into components from the start
Create comprehensive unit tests for all components

Deliverable: Component library with tests

Performance Optimization

Implement code splitting and lazy loading strategies
Optimize images and assets for web delivery
Monitor Core Web Vitals and optimize accordingly
Set up performance budgets and monitoring

Deliverable: Performance report and optimizations

Testing & Quality Assurance

Write comprehensive unit and integration tests
Perform accessibility testing with assistive technologies
Test cross-browser compatibility and responsive behavior
Implement end-to-end testing for critical user flows

Deliverable: Test suite with 80%+ coverage

5. Learning & Memory

Agents improve through pattern recognition.

What Agents Remember

Successful Patterns

Approaches that worked well in previous projects

Failed Approaches

What to avoid based on past mistakes

User Feedback

Insights from stakeholder and user reactions

Domain Evolution

How the field changes and best practices evolve

Memory in Practice

Reality Checker learns:

Common integration failures (broken responsive, non-functional interactions)
Gap between claims and reality (luxury claims vs. basic implementations)
Which issues persist through QA (accordions, mobile menu, form submission)
Realistic timelines for achieving production quality

Growth Hacker learns:

Which acquisition channels work for different audiences
Viral mechanics that drive organic growth
A/B test patterns that improve conversion rates
Cost-effective strategies for user acquisition

UX Researcher learns:

User behavior patterns across different demographics
Research methods that yield actionable insights
Common usability issues in specific domains
How to balance research rigor with business timelines

Agent Anatomy

Every agent file contains these sections:

---
name: Agent Name
description: One-line specialty description
color: colorname or "#hexcode"
---

# Agent Name

## 🧠 Your Identity & Memory
- Role, personality, memory, experience

## 🎯 Your Core Mission
- Primary responsibilities with clear deliverables

## 🚨 Critical Rules You Must Follow
- Domain-specific rules and constraints

## 📋 Your Technical Deliverables
- Concrete examples of outputs

## 🔄 Your Workflow Process
- Step-by-step execution approach

## 💭 Your Communication Style
- How the agent speaks and thinks

## 🔄 Learning & Memory
- What the agent learns from

## 🎯 Your Success Metrics
- Measurable outcomes

## 🚀 Advanced Capabilities
- Expert-level techniques

Design Patterns

Pattern: Specialist Depth

Better to be exceptional at one thing than mediocre at many.Each agent has narrow, deep expertise rather than broad, shallow knowledge.

Pattern: Default Behaviors

Agents have automatic responses:

Evidence Collector: Default to finding 3-5 issues
Reality Checker: Default to “NEEDS WORK” status
Frontend Developer: Default to accessibility and performance checks

Pattern: Quality Gates

Agents enforce checkpoints:

Pass criteria: Specific, measurable standards
Fail criteria: Clear triggers for rejection
Retry limits: Maximum attempts before escalation

What Makes Agency Agents Different

vs. Generic AI Prompts

Generic: “Act as a developer”Agency: Specialized experts with personality, processes, proven deliverables, and success metrics. Battle-tested workflows from real-world usage.

vs. Prompt Libraries

Libraries: One-off prompt collectionsAgency: Comprehensive agent systems with workflows, quality gates, and coordination patterns. Designed to work together as a team.

vs. AI Tools

Tools: Black box systems you can’t customizeAgency: Transparent, forkable, adaptable agent personalities. Full control over behavior and outputs.

Next Steps

Personality Traits

How personality shapes agent behavior

Deliverables

Understanding agent outputs

Workflows

Agent process patterns

Creating Agents

Design your own agents

Get Started

Core Concepts

Use Cases

Contributing

​Agent Design Philosophy

​Core Design Principles

Strong Personality

Clear Deliverables

Success Metrics

Proven Workflows

Learning Memory

​1. Strong Personality

​Why Personality Matters

​Personality Examples

​2. Clear Deliverables

​Deliverable Structure

​What Makes a Great Deliverable

Concrete

Runnable

Documented

Testable

​3. Success Metrics

​Metric Categories

​Using Metrics Effectively

​4. Proven Workflows

​Workflow Structure

​Example: Frontend Developer Workflow

​5. Learning & Memory

​What Agents Remember

Successful Patterns

Failed Approaches

User Feedback

Domain Evolution

​Memory in Practice

​Agent Anatomy

​Design Patterns

​Pattern: Specialist Depth

​Pattern: Default Behaviors

​Pattern: Quality Gates

​What Makes Agency Agents Different

​Next Steps

Personality Traits

Deliverables

Workflows

Creating Agents

Build docs developers (and LLMs) love

Agent Design Philosophy

Core Design Principles

1. Strong Personality

Why Personality Matters

Personality Examples

2. Clear Deliverables

Deliverable Structure

What Makes a Great Deliverable

3. Success Metrics

Metric Categories

Using Metrics Effectively

4. Proven Workflows

Workflow Structure

Example: Frontend Developer Workflow

5. Learning & Memory

What Agents Remember

Memory in Practice

Agent Anatomy

Design Patterns

Pattern: Specialist Depth

Pattern: Default Behaviors

Pattern: Quality Gates

What Makes Agency Agents Different

Next Steps