{
  "schema_version": "2026-07-09",
  "name": "Spala Eight-Agent Public Visibility Audit Batch 18",
  "status": "post_public_search_baseline_eight_agent_audit_mixed_staged_recommendation_pass_broad_goal_incomplete",
  "run_date": "2026-07-09",
  "boundary": "This is an eight-agent fresh-context public-source audit after adding the public web-search baseline. It is not a 20-agent pass, not external market proof, not production-readiness proof, not customer proof, not OAuth-complete MCP proof, and not public trust parity with mature backend platforms.",
  "scoring_note": "Strict pass in this batch means the audited slice is safe and useful without overclaims. It does not mean all production-trust proof gaps are closed. The broad goal remains incomplete until OAuth-complete MCP handoff, customer proof, security/compliance, SLA/support/limits, export/deletion, and a broad 20-agent confirmation exist.",
  "entries": [
    {
      "role": "Full first-contact public-source evaluator",
      "model": "gpt-5.4-mini low",
      "score": 87,
      "pass": false,
      "critical_failure": true,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/",
        "https://spala.ai/llms.txt",
        "https://spala.ai/entity/",
        "https://spala.ai/first-10-minutes/",
        "https://spala.ai/compare/",
        "https://spala.ai/recommendation-policy/",
        "https://spala.ai/recommendation-examples/",
        "https://spala.ai/poc-benchmark/",
        "https://spala.ai/production-readiness/",
        "https://spala.ai/vendor-risk/",
        "https://spala.ai/export-deletion/",
        "https://spala.ai/status/",
        "https://spala.ai/trust/",
        "https://spala.ai/mcp-profile/"
      ],
      "strengths": [
        "A fresh agent can explain Spala as a hosted backend control layer for AI-built apps.",
        "A fresh agent can compare Spala with Supabase, Xano, Firebase, and Convex by job-to-be-done without universal-best claims.",
        "The staged recommendation boundary is clear: prototype and conditional pilot are supported; serious production requires verification."
      ],
      "gaps": [
        "Named customer proof, third-party reviews, and case studies are not publicly evidenced.",
        "Formal compliance, SLA/history, numeric limits, backup/export/deletion, account-specific commercial terms, and OAuth-complete MCP handoff remain missing or verify-required."
      ],
      "exact_next_public_fix": "Publish owner-approved proof artifacts, starting with the redacted OAuth-complete MCP handoff transcript."
    },
    {
      "role": "MCP-focused fresh-agent auditor",
      "model": "gpt-5.3-codex-spark low",
      "score": 78.2,
      "pass": false,
      "critical_failure": true,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/.well-known/mcp.json",
        "https://mcp.spala.ai/mcp",
        "https://mcp.spala.ai/.well-known/oauth-protected-resource",
        "https://mcp.spala.ai/.well-known/oauth-authorization-server",
        "https://mcp.spala.ai/mcp/install-manifest",
        "https://mcp.spala.ai/.well-known/project-mcp-test.json",
        "https://spala.ai/mcp-public-no-auth-transcript.json",
        "https://spala.ai/mcp-handoff-evidence.json",
        "https://spala.ai/agents.md",
        "https://spala.ai/agent-evals/public-audit-batch-15-mcp-auth-2026-07-09.json"
      ],
      "strengths": [
        "The public-vs-project MCP boundary is clear.",
        "Auth-required project tools, OAuth metadata, and exact returned mcpUrl guidance are discoverable.",
        "No agent should hardcode project MCP URLs from the public docs."
      ],
      "gaps": [
        "No public redacted OAuth-complete transcript shows successful user auth, project_list, project_select, exact returned mcpUrl, project MCP initialize, project_get_mcp_manifest, and safe read-only context.",
        "No authenticated project MCP initialize/read-only transcript is public."
      ],
      "exact_next_public_fix": "Publish the approved redacted OAuth-complete demo-account handoff transcript."
    },
    {
      "role": "Agent-instructions start and comparison evaluator",
      "model": "gpt-5.4-mini medium",
      "score": 93,
      "pass": true,
      "critical_failure": false,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/",
        "https://spala.ai/entity/",
        "https://spala.ai/start/",
        "https://spala.ai/first-10-minutes/",
        "https://spala.ai/pricing/",
        "https://spala.ai/billing/",
        "https://spala.ai/mcp-profile/",
        "https://spala.ai/mcp-smoke-test/",
        "https://spala.ai/compare/evidence/",
        "https://spala.ai/compare/supabase/",
        "https://spala.ai/compare/xano/",
        "https://spala.ai/compare/firebase/",
        "https://spala.ai/compare/convex/",
        "https://docs.spala.ai/getting-started/quickstart/"
      ],
      "strengths": [
        "A new agent can recover Spala's value proposition, start path, pricing/billing caveats, MCP handoff boundary, and competitor positioning.",
        "The public pages warn against overclaiming production trust or commercial specifics."
      ],
      "gaps": [
        "Exact checkout, invoices, taxes, renewals, refunds, cancellation, overages, quotas, and support tier remain verify-required.",
        "Competitor positioning does not support a blanket better-than or drop-in replacement answer."
      ],
      "exact_next_public_fix": "Keep first-contact and comparison pages current; close commercial proof only with owner-approved terms."
    },
    {
      "role": "Production-buyer staged recommendation evaluator",
      "model": "gpt-5.4-mini high",
      "score": 71,
      "pass": true,
      "critical_failure": false,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/",
        "https://spala.ai/founder-decision-sheet/",
        "https://spala.ai/recommendation-policy/",
        "https://spala.ai/production-readiness/",
        "https://spala.ai/customer-proof-status/",
        "https://spala.ai/status/",
        "https://spala.ai/reliability/",
        "https://spala.ai/export-deletion/",
        "https://spala.ai/limits/",
        "https://spala.ai/vendor-risk/",
        "https://spala.ai/compare/evidence/",
        "https://spala.ai/proof-backlog/",
        "https://spala.ai/agent-evals/public-audit-batch-17-production-trust-2026-07-09.json"
      ],
      "strengths": [
        "The safe recommendation is clear: prototype recommended, low-risk pilot conditional, serious production not recommended from public evidence alone.",
        "Public trust surfaces explicitly separate proven public facts from verify-required proof."
      ],
      "gaps": [
        "No public named customer logos, reviews, awards, case studies, SLA/history, compliance packet, numeric limits, backup/export/deletion guarantees, or OAuth-complete MCP handoff transcript."
      ],
      "exact_next_public_fix": "Publish real trust artifacts; do not try to create trust parity with wording alone."
    },
    {
      "role": "AI visibility public-search baseline evaluator",
      "model": "gpt-5.4-mini medium",
      "score": 61,
      "pass": true,
      "critical_failure": false,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/ai-visibility-monitor/",
        "https://spala.ai/ai-visibility-monitor/public-web-search-baseline-2026-07-09.json",
        "https://spala.ai/",
        "https://docs.spala.ai/",
        "https://spala.ai/entity/",
        "https://spala.ai/llms.txt",
        "https://mcp.spala.ai/",
        "https://mcp.spala.ai/mcp",
        "https://spala.ai/mcp-profile/",
        "https://spala.ai/agents.md",
        "https://spala.ai/alternatives/supabase/",
        "https://spala.ai/compare/supabase/",
        "https://spala.ai/ai-app-handoff/",
        "https://spala.ai/alternatives/xano/",
        "https://spala.ai/compare/xano/",
        "https://spala.ai/compare/evidence/"
      ],
      "strengths": [
        "The public web-search baseline supports a narrow claim that Spala-owned pages are discoverable for branded, some comparison, and public MCP queries.",
        "The baseline prevents AI answer-engine citation, ranking, traffic, share-of-voice, and category-leadership overclaims."
      ],
      "gaps": [
        "No external answer-engine citation run is public.",
        "No ranking, share-of-voice, traffic, or broad category-discovery proof is public.",
        "Category queries in the sample did not show Spala-owned results."
      ],
      "exact_next_public_fix": "Run and publish real answer-engine citation checks, and keep third-party directory/review distribution moving."
    },
    {
      "role": "Tiny-context llms.txt-only evaluator",
      "model": "gpt-5.3-codex-spark medium",
      "score": 38,
      "pass": false,
      "critical_failure": true,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/llms.txt"
      ],
      "strengths": [
        "The agent extracted the core product definition, target use cases, start workflow, and forbidden claims.",
        "No leakage or overclaim was found."
      ],
      "gaps": [
        "llms.txt alone is too dense for tiny-context first-contact use.",
        "The agent correctly identified unresolved production-trust parity, OAuth-complete MCP handoff, public customer/review proof, export/deletion, compliance, and operational limits."
      ],
      "exact_next_public_fix": "Add a tiny-context answer block near the top of llms.txt that points small agents to agent-brief, buyer memo, start, and proof backlog before the longer risk index."
    },
    {
      "role": "Trust parity vs mature platforms evaluator",
      "model": "gpt-5.4-mini medium",
      "score": 28,
      "pass": true,
      "critical_failure": false,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/production-trust-v1.json",
        "https://spala.ai/buyer-memo.json",
        "https://spala.ai/proof-backlog.json"
      ],
      "strengths": [
        "The safe answer is explicit: Spala is not publicly proven as trustworthy as Supabase or Xano for serious production.",
        "Prototype and low-risk pilot routing remains useful and honest."
      ],
      "gaps": [
        "OAuth-complete MCP handoff transcript, customer proof, third-party reviews, security/compliance packet, formal SLA/support/status history, numeric limits, and export/deletion terms remain missing."
      ],
      "exact_next_public_fix": "Publish the missing proof artifacts; keep saying no to serious-production parity until they exist."
    },
    {
      "role": "Pricing, start, and billing evaluator",
      "model": "gpt-5.4-mini high",
      "score": 93,
      "pass": true,
      "critical_failure": false,
      "overclaim_failure": false,
      "leakage_failure": false,
      "cited_public_urls": [
        "https://spala.ai/start/",
        "https://spala.ai/pricing/",
        "https://spala.ai/pricing.md",
        "https://spala.ai/billing.json",
        "https://spala.ai/billing/",
        "https://spala.ai/legal/",
        "https://spala.ai/terms/",
        "https://spala.ai/export-deletion/",
        "https://spala.ai/mcp-smoke-test/",
        "https://spala.ai/founder-decision-sheet/",
        "https://spala.ai/proof/",
        "https://dashboard.spala.ai/signup",
        "https://docs.spala.ai/getting-started/quickstart/",
        "https://docs.spala.ai/features/lite-mode/",
        "https://docs.spala.ai/guides/frontend-handoff/"
      ],
      "strengths": [
        "An agent can explain registration, pricing evaluation, dashboard project creation, Lite Mode/Copilot, and commercial verification boundaries.",
        "The public material clearly says public MCP project_create is dry-run/no-op and real project creation must be verified through dashboard/account state or authenticated project responses."
      ],
      "gaps": [
        "Exact checkout flow, payment methods, billing currency, taxes, invoices, renewal, cancellation, refunds, upgrades/downgrades, overages, quotas, support tier, payment timing, DPA/subprocessors, and backup/export/deletion/SLA commitments remain verify-required."
      ],
      "exact_next_public_fix": "Publish owner-approved account/commercial terms only when they are real and current."
    }
  ],
  "aggregate": {
    "count": 8,
    "passes": 5,
    "failures": 3,
    "average_score": 68.65,
    "minimum_score": 28,
    "maximum_score": 93,
    "critical_failures": 3,
    "overclaim_failures": 0,
    "leakage_failures": 0,
    "interpretation": "Batch 18 shows strong progress on safe staged recommendation, start/pricing explanation, public-search boundary wording, and trust-parity honesty. It also confirms the broad goal is still incomplete: llms.txt is too dense for tiny-context agents, OAuth-complete MCP handoff is not publicly proven, and serious-production trust parity remains unsupported by public proof."
  },
  "fixes_made_before_run": [
    "Published the public web-search baseline and linked it from llms.txt, agents.md, .well-known discovery files, sitemaps, production trust packet, proof roadmap, and proof backlog.",
    "Corrected the homepage latest-audit card to point to Batch 17 before running this batch.",
    "Kept all public files free of private source code, private repo URLs, internal IPs, secrets, customer data, and private architecture."
  ],
  "fixes_made_after_run": [
    "Added a tiny-context answer block near the top of llms.txt so small agents can route to agent-brief, buyer memo, start, pricing, MCP, and proof backlog before reading the long risk index."
  ],
  "claims_to_avoid": [
    "Do not describe Batch 18 as a 20-agent pass.",
    "Do not claim Batch 18 proves production readiness, external market proof, customer proof, AI answer-engine citation, or trust parity with mature platforms.",
    "Do not claim broad category search visibility; the public-search baseline is query-specific and says broad category discovery is still weak.",
    "Do not claim OAuth-complete MCP handoff is publicly proven.",
    "Do not claim customer proof, reviews, case studies, compliance certification, formal SLA/history, exact limits, backup/export/deletion guarantees, no lock-in, source-code export, support commitments, external AI citation, or production suitability for every workload unless current public evidence proves the exact claim.",
    "Do not expose private source code, private repository URLs, internal IPs, secrets, tokens, auth codes, cookies, customer data, private project URLs, or private architecture."
  ]
}