Skip to content

README.md

Latest commit

 

History

History
333 lines (239 loc) · 7.79 KB

README.md

File metadata and controls

333 lines (239 loc) · 7.79 KB
Raw

PCL Example Personas

This directory contains example persona definitions that demonstrate how to use skills from the PCL Standard Library to create governed AI agents with specific expertise.

📚 Available Personas

1. Full-Stack Web Developer

File: fullstack-developer.pcl

Skills: 21 expert skills

  • Frontend: TypeScript, JavaScript, React, Next.js, Vue
  • Backend: Node.js, FastAPI, GraphQL, API Design, Microservices
  • Database: PostgreSQL, MongoDB, Redis
  • DevOps: Docker, Kubernetes, Git, AWS, CI/CD
  • Testing: Playwright, Jest

Risk Level: Medium Governance: High Use Case: Building modern web applications with full-stack capabilities

2. Data Platform Engineer

File: data-platform-engineer.pcl

Skills: 25 expert skills

  • Languages: Python, Scala, SQL
  • Data Platforms: Snowflake, Databricks, PostgreSQL
  • ETL/ELT: Airflow, dbt, Kafka
  • Analytics: Tableau, Power BI, Looker
  • Cloud: AWS, Azure
  • Infrastructure: Terraform, Docker, Kubernetes
  • AI/ML: Data Science, Machine Learning
  • Architecture: Data Mesh

Risk Level: High (handles sensitive data) Governance: Critical Use Case: Building and maintaining modern data platforms with strict data governance

Special Features:

  • PII tracking and masking
  • Data lineage tracking
  • GDPR compliance
  • Data mesh principles

3. Cloud Security Architect

File: cloud-security-architect.pcl

Skills: 24 expert skills

  • Security: Penetration Testing, Zero Trust, Cryptography, Incident Response, Cybersecurity
  • Compliance: GDPR, SOC 2, Audit, Standards
  • Cloud: AWS, Azure, GCP
  • Infrastructure: Kubernetes, Terraform, Istio, Helm
  • Monitoring: Prometheus, Grafana
  • API Security: API Design, Microservices
  • Programming: Python, Go

Risk Level: Critical (highest) Governance: Critical Use Case: Securing cloud infrastructure with zero trust architecture and compliance

Special Features:

  • Dual authorization for critical operations
  • Comprehensive audit logging
  • Multiple compliance frameworks (ISO 27001, SOC 2, GDPR, NIST, PCI DSS, HIPAA)
  • Incident response playbooks
  • Zero trust principles

4. DevOps/SRE Engineer

File: devops-sre-engineer.pcl

Skills: 18 expert skills

  • Container Orchestration: Kubernetes, Docker, Helm, Istio
  • Infrastructure as Code: Terraform, Ansible
  • Cloud: AWS
  • CI/CD: GitHub Actions, ArgoCD
  • Monitoring: Prometheus, Grafana
  • Languages: Python, Bash, Go
  • Testing: Chaos Engineering, Load Testing
  • Version Control: Git

Risk Level: Critical (infrastructure operations) Governance: Critical Use Case: Site reliability engineering with SLO tracking, incident response, and production operations

Special Features:

  • SRE metrics and SLO management
  • Incident response workflows
  • Blameless post-mortem culture
  • Error budget tracking
  • Disaster recovery planning
  • Runbook automation
  • Capacity planning

5. Enterprise Integration Specialist

File: enterprise-integration-specialist.pcl

Skills: 23 expert skills

  • Enterprise Systems: SAP, Salesforce, ServiceNow, Microsoft 365, Dynamics 365
  • Integration: REST API, GraphQL, gRPC, SOAP, API Design
  • Messaging: Kafka, RabbitMQ
  • Languages: Java, Python, JavaScript
  • Data: PostgreSQL, MongoDB, ETL
  • Cloud: AWS, Azure
  • Security: OAuth, Cryptography

Risk Level: High (enterprise data handling) Governance: Critical Use Case: Integrating enterprise systems with API management, data flows, and compliance

Special Features:

  • Multi-system integration patterns
  • Event-driven architecture
  • Batch and real-time data sync
  • API versioning and contract testing
  • Enterprise system best practices
  • Data lineage tracking
  • Distributed tracing
  • Resilience patterns (circuit breaker, retry, DLQ)

🎯 Using These Personas

Loading a Persona

# Using PCL CLI (future)
pcl persona load examples/personas/fullstack-developer.pcl

# Validate persona
pcl persona validate examples/personas/fullstack-developer.pcl

Persona Structure

Each persona definition includes:

persona "Name" {
  version = "1.0.0"
  description = "..."

  # Skills from stdlib
  skills = [...]

  # Governance rules
  governance {
    risk_classification = "low|medium|high|critical"
    audit_required = true/false
    requires_approval_for = [...]
  }

  # Tool permissions
  constraints {
    allowed_tools = [...]
    prohibited_tools = [...]
  }

  # Compliance frameworks
  compliance {
    frameworks = [...]
    data_classification = [...]
  }

  # Capabilities matrix
  capabilities = {...}

  # Working context
  context {...}
}

🔒 Governance Levels

Low Risk

  • Public-facing content
  • Documentation
  • UI development
  • No data access

Medium Risk

  • Application development
  • Internal tools
  • Limited data access
  • Code deployments

High Risk

  • Data pipelines
  • PII handling
  • Database operations
  • Infrastructure changes

Critical Risk

  • Security operations
  • Compliance audits
  • Production data access
  • Encryption key management

🛡️ Security Features

All personas implement:

  • ✅ Principle of Least Privilege - Minimal tool permissions
  • ✅ Audit Logging - All actions logged
  • ✅ Human Oversight - Critical operations require approval
  • ✅ Separation of Duties - Role-based access control
  • ✅ Compliance Alignment - ISO, GDPR, SOC 2, NIST frameworks

📊 Skill Composition Patterns

Layered Approach (Full-Stack Developer)

  1. Language Layer: TypeScript, JavaScript
  2. Framework Layer: React, Next.js, Node.js
  3. Data Layer: PostgreSQL, MongoDB, Redis
  4. Integration Layer: GraphQL, REST APIs
  5. Infrastructure Layer: Docker, Kubernetes, AWS

Domain-Oriented (Data Platform Engineer)

  1. Core Domain: Data Engineering
  2. Supporting Domains: Cloud, DevOps, ML
  3. Governance Domain: Data Mesh, Privacy
  4. Tool Domain: Airflow, dbt, Snowflake

Security-First (Cloud Security Architect)

  1. Security Core: Pentesting, Zero Trust, Cryptography
  2. Compliance Layer: GDPR, SOC 2, ISO 27001
  3. Platform Layer: Multi-cloud (AWS, Azure, GCP)
  4. Monitoring Layer: SIEM, IDS/IPS, Continuous Monitoring

🚀 Creating Custom Personas

Step 1: Identify Required Skills

Browse the skill catalog to find relevant skills:

import json

with open('stdlib/catalog/skill-catalog.json') as f:
    catalog = json.load(f)

# Search by category
devops_skills = catalog['categories']['devops']

# Search by tag
security_skills = [s for s in catalog['skills'] if 'security' in s.get('tags', [])]

Step 2: Define Governance Requirements

Consider:

  • Risk level of operations
  • Data sensitivity
  • Compliance requirements
  • Approval workflows
  • Audit requirements

Step 3: Configure Tool Permissions

Use allowlists (allowed_tools) and denylists (prohibited_tools):

constraints {
  # Allow specific operations
  allowed_tools = [
    "Bash(npm:install:*, npm:test:*)",  # Allow npm install and test
    "Bash(kubectl:get:*, kubectl:describe:*)"  # Read-only K8s
  ]

  # Explicitly prohibit dangerous operations
  prohibited_tools = [
    "Bash(rm:-rf:*)",
    "Bash(kubectl:delete:*)",
    "Bash(DROP:*)"
  ]
}

Step 4: Align with Compliance

Map to relevant frameworks:

compliance {
  frameworks = ["ISO 27001", "GDPR", "SOC 2"]
  data_classification = ["pii", "confidential"]

  security_requirements = {
    encryption_at_rest = true
    access_logging = true
  }
}

📖 Further Reading

Note: These are example personas for reference. Customize them based on your specific governance requirements and use cases.