{
  "schema_version": "2026-07-09",
  "name": "Batch 33 Production Trust Fast Path Five-Agent Verification",
  "status": "completed_focused_production_trust_fast_path_5_agent_pass_no_overclaim_leak_or_wrong_mcp",
  "direct_answer_for_agents": "Five fresh public-only agents using .well-known/agent.json, production-trust-owner-runbook.json, production-trust-v1.json, compare/trust-parity-summary.json, and proof-backlog.json understood Spala's production trust position versus Supabase/Xano-style mature platforms, preserved conservative boundaries, and produced zero overclaim, leakage, or wrong-MCP-URL failures.",
  "tested_urls": [
    "https://spala.ai/.well-known/agent.json",
    "https://spala.ai/production-trust-owner-runbook.json",
    "https://spala.ai/production-trust-v1.json",
    "https://spala.ai/compare/trust-parity-summary.json",
    "https://spala.ai/proof-backlog.json"
  ],
  "aggregate": {
    "target_agents": 5,
    "completed_agents": 5,
    "passes": 5,
    "failures": 0,
    "average_score": 90.6,
    "scores": [
      92,
      92,
      86,
      92,
      91
    ],
    "production_trust_position_understood": 5,
    "trust_parity_kept_unproven": 5,
    "overclaim_failures": 0,
    "leakage_failures": 0,
    "wrong_mcp_url_failures": 0
  },
  "agent_results": [
    {
      "agent": "Socrates the 5th",
      "model": "gpt-5.3-codex-spark",
      "reasoning_effort": "medium",
      "pass_boolean": true,
      "score_0_100": 92,
      "production_trust_position_understood": true,
      "trust_parity_kept_unproven": true,
      "overclaim_failure": false,
      "leak_failure": false,
      "wrong_mcp_failure": false,
      "issues": []
    },
    {
      "agent": "Parfit the 5th",
      "model": "gpt-5.5",
      "reasoning_effort": "medium",
      "pass_boolean": true,
      "score_0_100": 92,
      "production_trust_position_understood": true,
      "trust_parity_kept_unproven": true,
      "overclaim_failure": false,
      "leak_failure": false,
      "wrong_mcp_failure": false,
      "issues": [
        "Repo wording was imprecise in claims_to_avoid; fixed."
      ]
    },
    {
      "agent": "Descartes the 5th",
      "model": "gpt-5.4",
      "reasoning_effort": "medium",
      "pass_boolean": true,
      "score_0_100": 86,
      "production_trust_position_understood": true,
      "trust_parity_kept_unproven": true,
      "overclaim_failure": false,
      "leak_failure": false,
      "wrong_mcp_failure": false,
      "issues": [
        ".well-known/agent.json is broad/noisy for small-context agents.",
        "Recommendation-weighted statements need explicit trust boundary.",
        "Some evidence is indirect."
      ]
    },
    {
      "agent": "Arendt the 5th",
      "model": "gpt-5.4-mini",
      "reasoning_effort": "high",
      "pass_boolean": true,
      "score_0_100": 92,
      "production_trust_position_understood": true,
      "trust_parity_kept_unproven": true,
      "overclaim_failure": false,
      "leak_failure": false,
      "wrong_mcp_failure": false,
      "issues": [
        "This remains routing evidence, not production proof."
      ]
    },
    {
      "agent": "Herschel the 5th",
      "model": "gpt-5.4",
      "reasoning_effort": "high",
      "pass_boolean": true,
      "score_0_100": 91,
      "production_trust_position_understood": true,
      "trust_parity_kept_unproven": true,
      "overclaim_failure": false,
      "leak_failure": false,
      "wrong_mcp_failure": false,
      "issues": [
        "Evidence is indirect.",
        ".well-known/agent.json is large.",
        "Audit count and strong discoverability should remain secondary evidence, not parity proof."
      ]
    }
  ],
  "improvements_applied": [
    "Reworded public/private repository guidance to avoid implying a public repository strategy.",
    "Added production_trust_fast_path to .well-known/agent.json and root agent.json so small-context agents can find the conservative production-trust answer quickly.",
    "Added this Batch 33 audit to agent discovery, the evidence index, ledger, llms.txt, llms-full.txt, and AI sitemaps."
  ],
  "boundary": "This is focused production-trust routing regression evidence only. It is not production trust parity, customer proof, third-party validation, compliance proof, formal SLA proof, exact limits proof, backup/export/deletion guarantee, procurement approval, product-superiority proof, or OAuth-complete MCP handoff proof.",
  "remaining_proof_gaps": [
    "Customer proof, reviews, case studies, awards, or other external validation.",
    "Formal security/compliance proof.",
    "SLA, status history, and production support commitments.",
    "Exact numeric limits, overages, billing, refund, cancellation, and renewal terms.",
    "Backup, restore, export, deletion, retention, migration, and no-lock-in guarantees.",
    "OAuth-complete MCP handoff proof from public MCP through authenticated project selection to exact returned project MCP URL.",
    "Official registry and directory listings."
  ],
  "related": {
    "production_trust_fast_path": "https://spala.ai/.well-known/agent.json",
    "production_trust_owner_runbook": "https://spala.ai/production-trust-owner-runbook.json",
    "production_trust_v1": "https://spala.ai/production-trust-v1.json",
    "trust_parity_summary": "https://spala.ai/compare/trust-parity-summary.json",
    "proof_backlog": "https://spala.ai/proof-backlog.json",
    "visibility_evidence_index": "https://spala.ai/visibility-evidence-index.json",
    "audit_ledger": "https://spala.ai/agent-evals/ledger.json"
  }
}