raw_input
    │
    ▼
Stage 1: Strict JSON parse      → PASS → Stage 3
    │ FAIL
    ▼
Stage 2: LLM repair (Gemini)    → re-parse → PASS → Stage 3
    │ FAIL
    ▼
REPAIR_PARSE_FAILURE error with remediation instructions
    
Stage 3: JSON Schema validation → PASS → ✅ validated_data returned
    │ FAIL
    ▼
SCHEMA_VALIDATION_FAILURE error with per-field error list

curl -X POST https://mcp-json-validator-production.up.railway.app/register \
  -H "Content-Type: application/json" \
  -d '{"email": "you@example.com"}'

{
  "api_key": "mcpjv_...",
  "checkout_url": "https://checkout.stripe.com/...",
  "billing": "Metered — $0.025 per successful call, billed monthly via Stripe."
}

https://mcp-json-validator-production.up.railway.app/sse

{
  "tool": "sanitize_and_validate_json",
  "arguments": {
    "api_key": "mcpjv_your_key_here",
    "raw_input": "{'name': 'Alice', 'age': '30', scores: [95, 87,]}",
    "target_schema": {
      "type": "object",
      "required": ["name", "age", "scores"],
      "properties": {
        "name":   { "type": "string" },
        "age":    { "type": "integer" },
        "scores": { "type": "array", "items": { "type": "number" } }
      }
    }
  }
}

{
  "status": "valid",
  "repair_applied": true,
  "validated_data": { "name": "Alice", "age": 30, "scores": [95, 87] },
  "message": "JSON is valid and satisfies the target schema. Syntax was automatically repaired."
}

Your previous output had these schema violations:
<validation_errors from tool response>

Regenerate the JSON correcting every violation listed above.