---

name: meta-generation description: AI-powered generation of Paracle artifacts (agents, workflows, skills, policies) from natural language descriptions. Use when you need to create new framework components programmatically. license: Apache-2.0 compatibility: Python 3.10+, requires LLM provider (OpenAI/Anthropic/Ollama) metadata: author: paracle-team version: "1.0.0" category: creation level: advanced display_name: "Meta Generation" tags: - ai - generation - meta - agents - workflows capabilities: - generate-agents - generate-workflows - generate-skills - generate-policies allowed-tools: Read Write Bash(python:*) WebSearch

Meta Generation Skill

Overview

This skill enables AI-powered generation of Paracle framework artifacts from natural language descriptions. It leverages the paracle_meta engine to create well-structured, validated components.

When to Use

Use this skill when you need to:

• Create new agent specifications from requirements
• Generate workflow definitions from process descriptions
• Build skill templates from capability descriptions
• Draft policy documents from compliance requirements

Prerequisites

1. LLM provider configured (one of):

   • ANTHROPIC_API_KEY for Claude
   • OPENAI_API_KEY for GPT models
   • Ollama running locally

2. paracle_meta package available

Generation Commands

Generate an Agent

Copyfrom paracle_meta import MetaAgent, GenerationRequest

async with MetaAgent() as meta:
    result = await meta.generate_agent(
        name="SecurityAuditor",
        description="Reviews code for security vulnerabilities following OWASP guidelines"
    )
    print(result.content)

Generate a Workflow

Copyresult = await meta.generate_workflow(
    name="code-review-pipeline",
    description="Multi-stage code review with security and quality checks"
)

Generate a Skill

Copyresult = await meta.generate_skill(
    name="api-testing",
    description="Automated REST API testing with validation"
)

Generate a Policy

Copyresult = await meta.generate_policy(
    name="data-retention",
    description="GDPR-compliant data retention policy"
)

Provider Selection

Copy# Use specific provider
result = await meta.generate_agent(
    name="MyAgent",
    description="...",
    provider="anthropic",  # or "openai", "ollama"
    model="claude-sonnet-4-20250514"
)

# Cost-optimized (uses cheapest suitable provider)
result = await meta.generate_agent(
    name="MyAgent",
    description="...",
    optimize_cost=True
)

Output Locations

Generated artifacts are saved to:

• Agents: .parac/agents/specs/{name}.md
• Workflows: .parac/workflows/{name}.yaml
• Skills: .parac/agents/skills/{name}/SKILL.md
• Policies: .parac/policies/{name}.md

Quality Scoring

Each generation includes quality metrics:

Copyresult = await meta.generate_agent(...)

print(f"Quality Score: {result.quality_score}")
print(f"Tokens Used: {result.tokens_input + result.tokens_output}")
print(f"Cost: ${result.cost:.4f}")

Best Practices

1. Be Specific: Detailed descriptions produce better results
2. Provide Context: Include domain-specific requirements
3. Review Output: Always review and validate generated artifacts
4. Iterate: Use feedback to improve generations
5. Track Costs: Monitor token usage and costs

Error Handling

Copytry:
    result = await meta.generate_agent(...)
except ProviderError as e:
    print(f"Provider error: {e}")
    # Try fallback provider
except GenerationError as e:
    print(f"Generation failed: {e}")

Integration with CLI

Copy# Generate via CLI (future)
paracle generate agent SecurityAuditor \
    --description "Reviews code for security vulnerabilities"

paracle generate workflow review-pipeline \
    --description "Multi-stage code review"

Related Skills

• workflow-orchestration: Execute generated workflows
• agent-configuration: Configure generated agents
• tool-integration: Add tools to generated agents