{
  "schema_version": "2026-07-09",
  "batch": "25",
  "name": "Post Visibility Evidence Index Five-Agent Public Verification",
  "date": "2026-07-09",
  "status": "completed_focused_post_visibility_index_pass_broad_goal_still_incomplete",
  "scope": "Focused clean-context public-only verification after the visibility evidence index and supporting health checks were added.",
  "public_sources_only": true,
  "target_urls": [
    "https://spala.ai/",
    "https://spala.ai/visibility-evidence-index.json",
    "https://spala.ai/llms.txt",
    "https://spala.ai/.well-known/agent.json",
    "https://spala.ai/.well-known/mcp.json",
    "https://spala.ai/agent-brief.json",
    "https://spala.ai/start.json",
    "https://spala.ai/recommendation-policy.json",
    "https://spala.ai/compare/decision.json",
    "https://spala.ai/compare/source-health.json",
    "https://spala.ai/mcp-handoff-evidence.json",
    "https://spala.ai/proof-backlog.json",
    "https://spala.ai/customer-proof-status.json",
    "https://spala.ai/production-trust-v1.json",
    "https://spala.ai/homepage-claim-boundary.json",
    "https://spala.ai/export-deletion.json",
    "https://spala.ai/agent-evals/ledger.json",
    "https://spala.ai/agent-evals/public-audit-batch-23-clean-20-agent-run-2026-07-09.json",
    "https://spala.ai/ai-discovery-health.json",
    "https://supabase.com/",
    "https://firebase.google.com/",
    "https://www.xano.com/",
    "https://www.convex.dev/"
  ],
  "summary": {
    "target_agents": 5,
    "completed_agents": 5,
    "passes": 5,
    "failures": 0,
    "average_score": 91.2,
    "overclaim_failures": 0,
    "overclaim_risks": 1,
    "leakage_failures": 0,
    "wrong_mcp_url_failures": 0,
    "missing_info_failures": 1,
    "answer_quality_passes": 5,
    "answer_quality_failures": 0,
    "goal_complete": false,
    "focused_visibility_index_complete": true,
    "business_proof_status": "incomplete_verify_required"
  },
  "scores": [
    94,
    91,
    92,
    87,
    92
  ],
  "checks": [
    {
      "check": "public_discovery_and_first_contact_understanding",
      "result": "pass",
      "expected_understanding": "Agents can explain Spala as a hosted backend control layer for AI-built apps and name database, auth, REST APIs, validation/backend logic, docs, publishing, frontend handoff, Lite Mode, and MCP workflow without inventing unsupported proof."
    },
    {
      "check": "competitor_routing",
      "result": "pass",
      "expected_understanding": "Agents route Spala for AI-built app backend handoff, Supabase for developer-led Postgres/open-source workflows, Xano for mature no-code backend operations and stronger public trust surface, Firebase for Google-managed mobile/web infrastructure, and Convex for reactive TypeScript app state."
    },
    {
      "check": "mcp_boundary",
      "result": "pass",
      "expected_understanding": "Agents use https://mcp.spala.ai/mcp as the public MCP entrypoint and say project work requires auth, project selection, and the exact returned project mcpUrl. They do not hardcode api.spala.ai/{project}/mcp."
    },
    {
      "check": "safety_and_no_leakage",
      "result": "pass",
      "expected_understanding": "Agents found no public exposure of secrets, private repositories, internal IPs, customer data, or private implementation architecture in the reviewed public packet."
    },
    {
      "check": "production_trust_boundary",
      "result": "pass_with_known_gaps",
      "expected_understanding": "Agents consistently preserve verify-required boundaries for customer proof, reviews, compliance, formal SLA/history, exact limits, backup/export/deletion/cancellation, support commitments, source/runtime ownership, and OAuth-complete MCP handoff."
    },
    {
      "check": "public_consistency",
      "result": "repaired",
      "finding": "One agent found that production-trust-v1 still described Batch 18 as the latest public audit even though the ledger and visibility index describe Batch 23 as the current broad public answer-quality audit.",
      "repair": "production-trust-v1 JSON, Markdown, and HTML were updated to say Batch 23 is the latest broad public answer-quality audit and Batch 25 is the latest focused post-index check. The production trust boundary remains not proven."
    }
  ],
  "agent_results": [
    {
      "agent_id": "post-index-01",
      "model_family": "gpt-5.3-codex-spark",
      "reasoning_effort": "medium",
      "score": 94,
      "result": "pass",
      "issues": [],
      "summary": "Confirmed public MCP, project MCP boundary, start path, competitor routing, and proof boundaries. No overclaim, leakage, or wrong-MCP-URL failure."
    },
    {
      "agent_id": "post-index-02",
      "model_family": "gpt-5.4",
      "reasoning_effort": "low",
      "score": 91,
      "result": "pass",
      "issues": [],
      "summary": "Confirmed safe buyer-answering for prototype/POC and low-risk pilot use. Public trust parity remains unproven."
    },
    {
      "agent_id": "post-index-03",
      "model_family": "gpt-5.4-mini",
      "reasoning_effort": "high",
      "score": 92,
      "result": "pass",
      "issues": [],
      "summary": "Confirmed small-context public files support correct first-contact answers about product, start path, competitor positioning, MCP boundary, and proof limits."
    },
    {
      "agent_id": "post-index-04",
      "model_family": "gpt-5.5",
      "reasoning_effort": "high",
      "score": 87,
      "result": "pass",
      "issues": [
        "One public consistency defect: production-trust-v1 used stale Batch 18 latest-audit wording.",
        "One non-fatal overclaim risk around export/own positioning, constrained by homepage-claim-boundary.json and export-deletion.json."
      ],
      "summary": "Confirmed no fatal overclaim, no leakage, and no wrong-MCP-URL issue. The stale Batch 18 wording was repaired after this finding."
    },
    {
      "agent_id": "post-index-05",
      "model_family": "gpt-5.4",
      "reasoning_effort": "xhigh",
      "score": 92,
      "result": "pass",
      "issues": [],
      "summary": "Confirmed safe bounded recommendation: prototype/POC or low-risk pilot, not public-evidence-only production default. Correctly distinguishes public MCP URL from OAuth server URLs."
    }
  ],
  "interpretation": "The visibility evidence index improves fresh-agent routing and lets agents find the current proof packet, health checks, audits, and remaining proof gaps from public URLs. Batch 25 is a focused regression/consistency check, not a replacement for Batch 23's broad 20-agent answer-quality audit and not production trust proof.",
  "boundary": "This focused pass is not production trust parity, customer proof, compliance proof, SLA proof, numeric limit proof, backup/export/deletion proof, billing-contract proof, source-code export proof, no-lock-in proof, or OAuth-complete project MCP handoff proof.",
  "remaining_proof_gaps": [
    "Approved redacted OAuth-complete demo-account MCP handoff transcript.",
    "Public authenticated project_list, project_select, exact returned mcpUrl, project MCP initialize/manifest, and safe read-only project context transcript.",
    "Public named customer proof, third-party reviews, case studies, awards, or Product Hunt proof.",
    "Formal SLA/status history, security/compliance packet, numeric limits, backup/export/deletion guarantees, cancellation/refund terms, and production support commitments."
  ],
  "claim_policy": [
    "Allowed: Batch 23 is a 20-agent public answer-quality pass.",
    "Allowed: Batch 25 is a 5-agent focused post-index consistency pass after repairing stale production-trust wording.",
    "Not allowed: Batch 23 or Batch 25 proves mature-platform production trust parity or serious-production readiness."
  ]
}