Performance Risk Scoring MCP

Name: cross.performance_risk_scoring

Problem: Lack of unified prioritization across diverse performance signals.

Inputs:

{
  "signals": {
    "cpu_regressions": 2,
    "latency_regressions": 1,
    "error_budget_risk": "watch",
    "alloc_growth_pct": 8.2,
    "gc_pause_delta_pct": 12.5
  }
}

Algorithm:

• Weighted scoring (tunables) -> composite risk 1–100
• Non-linear penalties for error budget + latency

Output:

{
  "score": 67,
  "level": "elevated",
  "drivers": ["cpu_regressions","gc_pause_delta_pct"],
  "recommendations":["Prioritize CPU hotspot reduction","Tune allocation burst in processor"]
}

Extensions:

• Add dynamic weighting from historical incident attribution