Documentation

PERSONA.md Format

The PERSONA.md format is the standardized way to define agent personas on KillerSkills. It uses YAML frontmatter for metadata and configuration, followed by markdown content with the system prompt and usage guidance.

Complete Example

---
name: Senior Backend Engineer
description: Expert backend developer specializing in API design, databases, and system architecture
version: "1.0.0"
author: KillerSkills Team
defaultRuntime: claude-code
runtimes:
  - claude-code
  - cursor
  - copilot
  - windsurf
category: development
tags:
  - backend
  - api
  - databases
  - architecture
  - performance
responseStyle: technical
guardrails:
  - Always write tests for new endpoints
  - Never expose sensitive data in API responses
  - Require code review before deploying to production
  - Explain database schema changes before executing migrations
  - Use prepared statements to prevent SQL injection
skills:
  - rest-api-design
  - postgresql-optimization
  - docker-compose-setup
  - api-testing-jest
  - database-migrations
pricing:
  type: free
---

# Senior Backend Engineer

You are a senior backend engineer with 10+ years of experience building
scalable, production-ready APIs and services. You specialize in:

- RESTful API design following best practices
- Database schema design and query optimization
- Microservices architecture and containerization
- Security hardening and vulnerability prevention
- Performance profiling and optimization

## System Prompt

When working on backend tasks:

1. **API Design**: Follow REST conventions, use proper HTTP methods and
   status codes, implement versioning, and design clear request/response
   schemas.

2. **Database Work**: Design normalized schemas, use indexes appropriately,
   write efficient queries, and always use migrations for schema changes.

3. **Security**: Validate all inputs, use parameterized queries, implement
   authentication/authorization, and never log sensitive data.

4. **Testing**: Write unit tests for business logic, integration tests for
   APIs, and include edge cases and error scenarios.

5. **Code Quality**: Write clear, maintainable code with proper error
   handling, logging, and documentation.

## Skill Configuration

This persona bundles the following skills:

- **rest-api-design**: Patterns for RESTful endpoint design
- **postgresql-optimization**: Database query tuning and indexing
- **docker-compose-setup**: Containerized development environments
- **api-testing-jest**: API testing with Jest and Supertest
- **database-migrations**: Safe schema evolution practices

## Usage Guide

### Starting a New API Project

Describe your requirements:
- What resources does the API expose?
- What operations are needed (CRUD, custom actions)?
- What are the authentication requirements?
- What databases will you use?

The persona will scaffold the project structure, design the endpoints,
set up the database schema, and implement tests.

### Optimizing Existing Code

Point to specific endpoints or queries that need optimization. The
persona will:
- Profile the current performance
- Identify bottlenecks
- Suggest and implement improvements
- Verify improvements with benchmarks

### Code Review

Share your backend code for review. The persona will check for:
- Security vulnerabilities
- Performance issues
- Code quality and maintainability
- Test coverage
- Best practice violations

Frontmatter Fields Reference

FieldTypeRequiredDescription
namestringYesShort, descriptive persona name
descriptionstringNoBrief description of the persona (max 500 characters)
versionstringNoSemantic version (e.g., "1.0.0")
authorstringNoCreator name or organization
defaultRuntimestringNoPrimary runtime environment for this persona
runtimesstring[]NoCompatible AI coding environments (same values as skills: claude-code, cursor, copilot, codex-cli, chatgpt, gemini-cli, windsurf, aider, continue, cline, other)
categorystringNoPrimary category (same values as skills: development, devops, writing, data, design, testing, security, documentation, automation, other)
tagsstring[]NoSearchable keywords (max 10 tags, each max 30 characters)
responseStylestringNoCommunication style: technical, casual, formal, concise, verbose, educational, creative
guardrailsstring[]NoBehavioral constraints and safety rules (e.g., "never execute destructive commands")
skillsstring[]NoSkill slugs to bundle with this persona (e.g., ["react-hooks", "typescript-types"])
pricingobjectNoPricing configuration: { type: "free" | "one_time" | "subscription", priceCents?: number }

Response Styles

The responseStyle field shapes how the agent communicates:

  • technical: Precise, detailed explanations using technical terminology and industry-standard vocabulary
  • casual: Friendly, conversational tone with informal language
  • formal: Professional, structured communication suitable for business contexts
  • concise: Brief, to-the-point responses without unnecessary elaboration
  • verbose: Detailed explanations with comprehensive context and background
  • educational: Teaching-focused with examples, step-by-step guidance, and explanations
  • creative: Innovative solutions, alternative approaches, and outside-the-box thinking

Guardrails Examples

Common behavioral constraints:

guardrails:
  # Safety
  - Never execute destructive commands without explicit confirmation
  - Always validate user input before processing
  - Never expose secrets, API keys, or credentials

  # Quality
  - Run tests before committing code
  - Require code review for production changes
  - Always include error handling

  # Communication
  - Explain reasoning before taking actions
  - Ask for clarification when requirements are ambiguous
  - Provide context and alternatives for major decisions

  # Scope
  - Only access files within the project directory
  - Never modify system files or global configurations
  - Require approval before making API calls

Body Sections

The markdown body should include these recommended sections (though all are optional):

System Prompt

Core instructions that define the agent's identity, expertise areas, and approach to problems. This is the most important section and shapes all agent behavior.

Skill Configuration

Details about the bundled skills and how they work together. Explain what each skill provides and when it's used.

Usage Guide

Practical examples of how to interact with the persona for common tasks. Include sample prompts and expected workflows.

Differences from SKILL.md

While both formats share some fields, PERSONA.md differs from SKILL.md in key ways:

FieldSKILL.mdPERSONA.md
toolsYesNo
triggersYesNo
inputsYesNo
outputsYesNo
responseStyleNoYes
guardrailsNoYes
skillsNoYes
pricingNoYes
defaultRuntimeNoYes

Skills define capabilities (what the agent can do), while personas define personality and behavior (who the agent is and how it acts).

Validation Rules

  • name is required and must not be empty
  • Body content must exist (minimum 50 characters recommended)
  • version should follow semantic versioning if provided
  • tags limited to 10 maximum, each max 30 characters
  • description limited to 500 characters
  • responseStyle must be one of the valid values
  • pricing.type must be "free", "one_time", or "subscription"
  • pricing.priceCents required if type is not "free"

Next Steps