PCL — Persona Control Language

╔═══════════════════════════════════════════════════════════════════════════════╗
║                                                                               ║
║   ██████╗  ██████╗██╗                                                         ║
║   ██╔══██╗██╔════╝██║         The World's First Programming Language          ║
║   ██████╔╝██║     ██║              for AI Persona Management                  ║
║   ██╔═══╝ ██║     ██║                                                         ║
║   ██║     ╚██████╗███████╗    Make AI behavior programmable, portable,        ║
║   ╚═╝      ╚═════╝╚══════╝           and predictable.                         ║
║                                                                               ║
╚═══════════════════════════════════════════════════════════════════════════════╝

Standards Compliance:

What is PCL?

PCL (Persona Control Language) is a governance-first programming language for AI persona management and multi-agent orchestration. Unlike traditional application languages, PCL is designed for accountability, security, and compliance in AI systems.

PCL = Terraform + OpenPolicyAgent + AI Personas

PCL enables enterprises and developers to:

Define personas with explicit capabilities, constraints, and risk classifications (ISO 42001)
Govern AI behavior through auditable policies and access controls (ISO 27001)
Orchestrate complex multi-agent workflows with human oversight
Deploy consistently across Claude, GPT, Gemini, Azure, and open-source LLMs
Audit every action with immutable logs aligned to compliance frameworks
Comply with EU AI Act, GDPR, OWASP LLM Top 10, and Zero Trust principles

Why PCL Exists

Traditional AI development lacks:

✗ Accountability – Who made what decision? ✗ Portability – Vendor lock-in ✗ Security – No defense against prompt injection, excessive agency ✗ Compliance – ISO, EU AI Act, OWASP alignment ✗ Governance – No clear policy enforcement

PCL solves this by treating AI personas as governed entities, not just code.

⚡ Quick Start - What Works Today

📖 New User? Read the Getting Started Guide (Current Features) for a practical, working introduction to PCL's implemented features.

What PCL can do right now (Phases 1 & 2.1 & 2.2 Complete):

Core Language (Phase 1) ✅

✅ Parse PCL files to AST
✅ Type checking and semantic analysis
✅ Runtime with 3 LLM providers (Mock, Claude, OpenAI)
✅ Registry system with 4 backends (Memory, JSON File, SQLite, PostgreSQL)
✅ Full-text search with filters
✅ CLI with 7 registry commands
✅ Expression evaluator

IDE Support (Phase 2.1) ✅ NEW!

✅ Full Language Server Protocol (LSP) implementation
✅ VSCode extension with syntax highlighting
✅ Real-time diagnostics and error detection
✅ IntelliSense auto-completion (26 keywords, 13 snippets)
✅ Hover documentation (30+ properties)
✅ Go to definition (Ctrl+Click)
✅ Find all references
✅ Document outline/symbols
✅ Auto-formatting

Skills Ecosystem (Phase 2.2) ✅ NEW!

✅ 100% compatible with Agent Skills specification
✅ 95% compatible with Claude Code SKILL.md format
✅ Import skills from agentskills.io and Claude Code
✅ Bidirectional skill conversion (PCL ↔ SKILL.md)
✅ Skill loader for YAML frontmatter + Markdown
✅ Progressive disclosure pattern
✅ Multi-file skills support

Example Persona (Current Syntax)

// Define a security analyst persona
pub persona SEC {
  intent: "Identify and mitigate security vulnerabilities"
  tone: vigilant

  skills {
    "OWASP Top 10"
    "STRIDE threat modeling"
    "Security code review"
  }

  constraints {
    "Always assume breach"
    maxResponseTime <= 5s
  }

  pub fn analyze(target: St

...

personamanagmentlayer/pcl

README