diamitani/rostr-agent

ROSTR Well-Architected Agent Skill

Description

Perform AWS Well-Architected Framework reviews on agent outputs. Evaluates against 6 pillars: Operational Excellence, Security, Reliability, Performance Efficiency, Cost Optimization, and Sustainability.

Trigger Conditions

• When an agent produces output that needs quality review
• Before deploying agent configurations
• During periodic architectural reviews
• When cost/security concerns are raised

Input Format

{
  "agent_output": {
    "result": "string",
    "metadata": {
      "model_used": "string",
      "tokens_used": {"prompt": int, "completion": int},
      "duration_ms": int
    },
    "logs": [...],
    "pii_detected": false,
    "graviton_used": true
  },
  "review_type": "full" | "security" | "cost" | "quick"
}

Output Format

{
  "SAR": {
    "SchemaVersion": "2020-07-01",
    "Review": {
      "OverallScore": 4.2,
      "Pillars": [...]
    },
    "Risks": {
      "High": [...],
      "Medium": [...],
      "Low": [...]
    },
    "ImprovementItems": [...]
  }
}

Usage

from skills.well_architected import WellArchitectedSkill

skill = WellArchitectedSkill()
review = await skill.review(agent_output, review_type="full")

# Get SAR formatted output
sar = skill.to_sar_format(review)

Tool Dependencies

• none

Example Prompts

Review this agent output for Well-Architected compliance

Perform security-focused review

Check cost optimization opportunities

Version

1.0.0