log-router-bq-report

name	log-router-bq-report
description	Sets up a Cloud Logging Log Router sink to stream live Cloud Run server logs into a BigQuery dataset, then queries BigQuery via MCP to generate a quick traffic and error report, and a game analytics report with feature insights. Use when asked to route Cloud Run logs to BigQuery, set up log export, analyze server logs, generate a log report from BigQuery, or surface feature ideas from game event data.
license	Proprietary
compatibility	Requires google-cloud-logging and google-bigquery MCP servers to be enabled in mcp_config.json. The sink creation step requires gcloud CLI (authenticated) to be available in the shell. Designed for Gemini/Antigravity agent with access to Bash and MCP tools.
metadata	{"author":"<SERVICE_NAME>-team","version":"1.0","service":"<SERVICE_NAME>","region":"us-central1"}
allowed-tools	mcp__google-cloud-logging mcp__google-bigquery Bash

Configuration

Edit these values to adapt this skill to your project.

Variable	Default	Description
SERVICE_NAME	dinoquest	Cloud Run service name
BQ_DATASET	dinoquest_logs	BigQuery dataset name for log export
LOG_SINK_NAME	dinoquest-bq-sink	Cloud Logging sink name

---

Overview

Before doing any work, output this pipeline flow diagram so the user can see what is coming:

  DinoQuest Log Router + BigQuery Report Pipeline
  ══════════════════════════════════════════════════

  ┌─────────────────────────────┐
  │  Resolve GCP Project        │
  └──────────────┬──────────────┘
                 │
                 ▼
  ┌─────────────────────────────┐
  │  Phase 1: Log Router Setup  │
  │  Check if sink exists       │
  └──────────────┬──────────────┘
                 │
         ┌───────┴────────┐
         ▼                ▼
    [✔] exists        [new]
    skip to           Create BQ dataset
    Phase 2           Create sink
                      Grant BQ permissions
         └───────┬────────┘
                 │
                 ▼
  ┌─────────────────────────────┐
  │  Phase 2: BQ Log Report     │
  │  Discover log table         │
  │  Query: traffic summary     │
  │  Query: top error messages  │
  │  Output report              │
  └──────────────┬──────────────┘
                 │
                 ▼
  ┌─────────────────────────────┐
  │  Phase 3: Feature Insights  │
  │  Win rate by dino type      │
  │  Coins per outcome          │
  │  Dino reuse rate            │
  │  Habitat × Diet → Type      │
  └──────────────┬──────────────┘
                 │
                 ▼
  ┌─────────────────────────────┐
  │  Generate HTML Report       │
  │  Save to reports/           │
  │  Print file path            │
  └─────────────────────────────┘

  ══════════════════════════════════════════════════

This skill has three phases:

1. Setup — Creates a Cloud Logging Log Router sink to continuously stream DinoQuest2 Cloud Run logs into a BigQuery dataset (idempotent: skipped if the sink already exists).
2. Report — Queries the BigQuery log table via MCP to produce a live traffic and error summary.
3. Feature Insights — Analyzes game event telemetry (DINO_CREATED, GAME_START, GAME_END) to surface data-driven product ideas around personalization, difficulty, retention, and progression.

See references/SETUP_NOTES.md for architecture details and IAM notes.

---

Phase 1: Set Up the Log Router Sink

Step 1.1 — Resolve Project ID

Try MCP first. Use the google-resource-manager MCP server to search for the active project:

Tool: google-resource-manager → search_projects

Extract the projectId from the first result that matches the DinoQuest project (look for dinoquest in the name or ID, or the project associated with the current credentials).

If the MCP call fails or returns no results, fall back to CLI:

gcloud config get-value project

Store the resolved value as PROJECT_ID for all subsequent steps.

---

Step 1.2 — Check if Sink Already Exists

Try MCP first. Use the google-bigquery MCP server to check whether the destination dataset already has log tables (which proves the sink has been flowing data):

Tool: google-bigquery → list_table_ids
Parameters:
  - project: <PROJECT_ID>
  - dataset: <BQ_DATASET>

• If the dataset exists and contains run_googleapis_com_requests_* tables → sink already exists, skip to Phase 2.
• If the MCP call errors or the dataset is missing → fall back to CLI to confirm:

bash scripts/check_sink.sh "$PROJECT_ID"

• If the sink <LOG_SINK_NAME> exists → skip to Phase 2.
• If it does not exist → continue to Step 1.3.

---

Step 1.3 — Create the BigQuery Dataset (if needed)

Use the google-bigquery MCP server to check whether the target dataset exists:

Tool: google-bigquery → list_dataset_ids
Parameters:
  - project: <PROJECT_ID>

If the dataset <BQ_DATASET> is not in the list, create it:

bash scripts/create_bq_dataset.sh "$PROJECT_ID"

---

Step 1.4 — Create the Log Router Sink

Run the sink creation script:

bash scripts/create_sink.sh "$PROJECT_ID"

This script:

• Creates a sink named <LOG_SINK_NAME>
• Filters to Cloud Run logs for the <SERVICE_NAME> service
• Destinations to bigquery.googleapis.com/projects/$PROJECT_ID/datasets/<BQ_DATASET>
• Captures the sink's writerIdentity output

---

Step 1.5 — Grant BigQuery Permissions to the Sink Writer

After the sink is created, the Log Router generates a writerIdentity service account. Grant it write access to the dataset:

bash scripts/grant_sink_permissions.sh "$PROJECT_ID"

This grants the writerIdentity the roles/bigquery.dataEditor role on the <BQ_DATASET> dataset so the sink can insert rows.

Note: Logs will begin flowing within ~1–2 minutes of the sink being created. If you run the report immediately after setup, results may be empty — wait a few minutes for data to populate, or proceed to Phase 2 to check existing data.

---

Phase 2: Generate the BigQuery Log Report

Step 2.1 — Discover the Log Table

Use the google-bigquery MCP server to confirm the log table exists in the dataset:

Tool: google-bigquery → list_table_ids
Parameters:
  - project: <PROJECT_ID>
  - dataset: <BQ_DATASET>

Cloud Logging creates tables named by date (e.g., run_googleapis_com_requests_YYYYMMDD). Identify the most recent table name for use in queries.

---

Step 2.2 — Query: Request Rate Summary (Last Hour)

Tool: google-bigquery → execute_sql
SQL:
  SELECT
    FORMAT_TIMESTAMP('%Y-%m-%d %H:%M', timestamp, 'UTC') AS minute,
    COUNT(*) AS request_count,
    COUNTIF(httpRequest.status >= 500) AS server_errors,
    COUNTIF(httpRequest.status >= 400 AND httpRequest.status < 500) AS client_errors,
    ROUND(AVG(httpRequest.latency * 1000), 0) AS avg_latency_ms
  FROM `<PROJECT_ID>.<BQ_DATASET>.<TABLE_NAME>`
  WHERE timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 24 HOUR)
  GROUP BY minute
  ORDER BY minute DESC
  LIMIT 30

---

Step 2.3 — Query: Top Error Messages (Last Hour)

Tool: google-bigquery → execute_sql
SQL:
  SELECT
    textPayload AS error_message,
    severity,
    COUNT(*) AS occurrences
  FROM `<PROJECT_ID>.<BQ_DATASET>.<TABLE_NAME>`
  WHERE timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 24 HOUR)
    AND severity IN ('ERROR', 'CRITICAL', 'ALERT', 'EMERGENCY')
  GROUP BY error_message, severity
  ORDER BY occurrences DESC
  LIMIT 10

---

Step 2.4 — Report Results

Output the following structured report:

=== DinoQuest2 Log Report (BigQuery) ===
Project:    <PROJECT_ID>
Dataset:    <BQ_DATASET>
Table:      <TABLE_NAME>
Report Time: <ISO-8601 timestamp>
Window:     Last 24 hours

--- Traffic Summary ---
[Table: minute | request_count | server_errors | client_errors | avg_latency_ms]

--- Top Errors ---
[Table: error_message | severity | occurrences]

--- Sink Status ---
Sink Name:       <LOG_SINK_NAME>
Created:         YES / already existed
Writer Identity: <writerIdentity email>

---

---

Phase 3: Game Feature Insights

Analyzes game events in run_googleapis_com_stdout_YYYYMMDD. The backend logs via print(json.dumps({...})), so events land in the textPayload column as plain JSON strings — there is no jsonPayload column. Extract fields with JSON_VALUE(textPayload, '$.field').

• Booleans (won, is_reuse): compare as strings — JSON_VALUE(textPayload, '$.won') = 'true'
• Numerics (score, coins, speed): SAFE_CAST(JSON_VALUE(textPayload, '$.score') AS FLOAT64)
• Filter for events: WHERE JSON_VALUE(textPayload, '$.event') = 'GAME_END'

Run all four queries in parallel. Use the most recent stdout table (same date suffix as Phase 2).

---

Step 3.1 — Query: Win Rate by Dino Type (Difficulty Tuning)

If Agile win rate is significantly higher than other types, rebalancing is needed.

SELECT
  JSON_VALUE(textPayload, '$.dino_type') AS dino_type,
  COUNT(*) AS total_games,
  COUNTIF(JSON_VALUE(textPayload, '$.won') = 'true') AS wins,
  ROUND(COUNTIF(JSON_VALUE(textPayload, '$.won') = 'true') / COUNT(*) * 100, 1) AS win_rate_pct,
  ROUND(AVG(SAFE_CAST(JSON_VALUE(textPayload, '$.score') AS FLOAT64)), 0) AS avg_score,
  ROUND(AVG(SAFE_CAST(JSON_VALUE(textPayload, '$.speed') AS FLOAT64)), 2) AS avg_speed
FROM `<PROJECT_ID>.<BQ_DATASET>.<STDOUT_TABLE>`
WHERE JSON_VALUE(textPayload, '$.event') = 'GAME_END'
  AND JSON_VALUE(textPayload, '$.dino_type') IS NOT NULL
GROUP BY dino_type
ORDER BY win_rate_pct DESC

Signal: win_rate_pct > 70% for a single type → rebalance that type's speed/stats.

---

Step 3.2 — Query: Coins per Outcome (Progression Currency)

Low coin yield on wins → no incentive to keep playing.

SELECT
  JSON_VALUE(textPayload, '$.won') = 'true' AS won,
  COUNT(*) AS games,
  ROUND(AVG(SAFE_CAST(JSON_VALUE(textPayload, '$.coins') AS FLOAT64)), 2) AS avg_coins,
  MIN(SAFE_CAST(JSON_VALUE(textPayload, '$.coins') AS FLOAT64)) AS min_coins,
  MAX(SAFE_CAST(JSON_VALUE(textPayload, '$.coins') AS FLOAT64)) AS max_coins
FROM `<PROJECT_ID>.<BQ_DATASET>.<STDOUT_TABLE>`
WHERE JSON_VALUE(textPayload, '$.event') = 'GAME_END'
GROUP BY won

Signal: avg_coins on win ≤ 3 → consider a coin boost or streak bonus to drive retention.

---

Step 3.3 — Query: Dino Reuse Rate (Personalization Signal)

High reuse of specific dinos reveals player favorites and attachment patterns.

SELECT
  JSON_VALUE(textPayload, '$.dino_name') AS dino_name,
  JSON_VALUE(textPayload, '$.dino_type') AS dino_type,
  COUNT(*) AS total_starts,
  COUNTIF(JSON_VALUE(textPayload, '$.is_reuse') = 'true') AS reuse_starts,
  COUNTIF(JSON_VALUE(textPayload, '$.is_reuse') = 'false') AS fresh_starts
FROM `<PROJECT_ID>.<BQ_DATASET>.<STDOUT_TABLE>`
WHERE JSON_VALUE(textPayload, '$.event') = 'GAME_START'
  AND JSON_VALUE(textPayload, '$.dino_name') IS NOT NULL
GROUP BY dino_name, dino_type
ORDER BY reuse_starts DESC
LIMIT 10

Signal: Dinos with reuse_starts > 3 → show "your dino" personality card to returning players.

---

Step 3.4 — Query: Habitat + Diet → Dino Type Correlation (Personalization Engine)

Reveals whether habitat/diet choices reliably predict dino type — enables previewing type before confirm.

SELECT
  JSON_VALUE(textPayload, '$.habitat') AS habitat,
  JSON_VALUE(textPayload, '$.diet') AS diet,
  JSON_VALUE(textPayload, '$.generated_type') AS dino_type,
  COUNT(*) AS occurrences
FROM `<PROJECT_ID>.<BQ_DATASET>.<STDOUT_TABLE>`
WHERE JSON_VALUE(textPayload, '$.event') = 'DINO_CREATED'
GROUP BY habitat, diet, dino_type
ORDER BY occurrences DESC
LIMIT 20

Signal: A habitat+diet combo that always maps to one type (100% of rows) is deterministic enough to show a type preview before the user confirms.

Note: DINO_CREATED only fires when users generate a new dino via /api/generate. If the table was just created today, this may return empty. Always use the 7-day wildcard:

FROM `<PROJECT_ID>.<BQ_DATASET>.run_googleapis_com_stdout_*`
WHERE _TABLE_SUFFIX >= FORMAT_DATE('%Y%m%d', DATE_SUB(CURRENT_DATE(), INTERVAL 7 DAY))
  AND JSON_VALUE(textPayload, '$.event') = 'DINO_CREATED'

---

Step 3.5 — Generate Interactive HTML Report

Instead of printing plain text, generate a standalone animated HTML file and save it to reports/dinoquest-insights-<YYYYMMDD-HHmmss>.html (create the reports/ directory if it doesn't exist). After writing the file, print:

Report saved → reports/dinoquest-insights-<timestamp>.html
Open it in any browser to view the interactive dashboard.

---

HTML template

Write the file using the exact structure below. Replace every /* DATA */ placeholder with real values from the query results. The file must be fully self-contained (no external files — Chart.js and Google Fonts are loaded from CDN).

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8"/>
<meta name="viewport" content="width=device-width,initial-scale=1"/>
<title>DinoQuest Insights</title>
<link href="https://fonts.googleapis.com/css2?family=Playfair+Display:ital,wght@0,500;0,700;1,500&family=Nunito:wght@400;600;700;900&display=swap" rel="stylesheet"/>
<script src="https://cdn.jsdelivr.net/npm/chart.js@4.4.0/dist/chart.umd.min.js"></script>
<style>
  :root {
    --green:#15803d; --green-light:#f0fdf4; --green-dark:#14532d;
    --yellow:#d97706; --yellow-dark:#b45309; --orange:#c2410c;
    --red:#b91c1c; --blue:#1e40af; --purple:#7c3aed; --pink:#be185d;
    --bg:#fefce8; --card:#fffbeb; --ink:#292524; --muted:#78716c;
    --rule:#e7e5e4;
    --shadow:0 4px 16px rgba(20,83,45,.10);
  }
  *{box-sizing:border-box;margin:0;padding:0;}
  body{background:var(--bg);font-family:'Nunito',sans-serif;color:var(--ink);overflow-x:hidden;}

  /* ── editorial header ── */
  header{
    position:relative;z-index:1;text-align:center;padding:3.5rem 1rem 2.5rem;
    background:linear-gradient(180deg,#14532d 0%,#166534 100%);
    border-bottom:1px solid var(--yellow-dark);
  }
  header::before{
    content:'';position:absolute;inset:0;
    background:radial-gradient(circle at 50% 0%,rgba(217,119,6,.10) 0%,transparent 60%);
  }
  .header-eyebrow{
    position:relative;z-index:1;color:#fde68a;font-family:'Playfair Display',serif;
    font-style:italic;font-weight:500;font-size:.9rem;letter-spacing:.28em;
    text-transform:uppercase;opacity:.9;margin-bottom:.9rem;
  }
  .header-eyebrow::before,.header-eyebrow::after{
    content:'';display:inline-block;width:2.4rem;height:1px;background:#fde68a;
    vertical-align:middle;margin:0 .9rem;opacity:.55;
  }
  .header-title{
    position:relative;z-index:1;font-family:'Playfair Display',serif;font-weight:700;
    font-size:clamp(2.2rem,4.8vw,3.6rem);color:#fefce8;
    letter-spacing:.005em;line-height:1.1;
    animation:fadeUp .7s ease-out both;
  }
  @keyframes fadeUp{from{opacity:0;transform:translateY(10px);}to{opacity:1;transform:translateY(0);}}
  .header-rule{
    position:relative;z-index:1;width:64px;height:2px;margin:1rem auto 0;
    background:var(--yellow);opacity:.85;
  }
  .header-sub{
    position:relative;z-index:1;color:#dcfce7;font-family:'Playfair Display',serif;
    font-style:italic;font-size:1rem;font-weight:500;margin-top:.9rem;opacity:.85;
  }
  .header-badge{
    position:relative;z-index:1;display:inline-block;background:transparent;color:#fde68a;
    font-family:'Nunito',sans-serif;font-weight:700;font-size:.72rem;
    padding:.4rem 1rem;border-radius:9999px;margin-top:1.4rem;
    border:1px solid rgba(253,230,138,.45);letter-spacing:.18em;text-transform:uppercase;
  }

  /* ── main grid ── */
  main{position:relative;z-index:1;max-width:1200px;margin:0 auto;padding:2.5rem 1rem 4rem;}
  .section-title{
    font-family:'Playfair Display',serif;font-weight:700;font-size:1.55rem;
    color:var(--green-dark);margin:2.75rem 0 1.25rem;
    display:flex;align-items:center;gap:.75rem;
    border-bottom:1px solid var(--rule);padding-bottom:.6rem;
  }
  .section-title .pill{
    font-family:'Nunito',sans-serif;font-size:.68rem;font-weight:700;
    background:transparent;color:var(--yellow-dark);padding:.2rem .65rem;
    border-radius:9999px;letter-spacing:.18em;text-transform:uppercase;
    border:1px solid var(--yellow-dark);
  }

  /* ── cards ── */
  .card{
    background:var(--card);border-radius:.5rem;
    border:1px solid var(--rule);box-shadow:var(--shadow);
    padding:1.5rem;transition:transform .2s,box-shadow .2s;
    animation:cardIn .5s ease both;
  }
  .card:hover{transform:translateY(-2px);box-shadow:0 8px 24px rgba(20,83,45,.14);}
  @keyframes cardIn{from{opacity:0;transform:translateY(12px);}to{opacity:1;transform:translateY(0);}}
  .cards-grid{display:grid;grid-template-columns:repeat(auto-fit,minmax(220px,1fr));gap:1.25rem;margin-bottom:1.5rem;}

  /* ── stat bubble ── */
  .stat-bubble{text-align:center;}
  .stat-bubble .value{
    font-family:'Playfair Display',serif;font-weight:700;
    font-size:2.6rem;line-height:1;color:var(--green-dark);
  }
  .stat-bubble .label{font-family:'Playfair Display',serif;font-size:.78rem;font-weight:500;color:var(--muted);letter-spacing:.22em;text-transform:uppercase;margin-top:.55rem;}
  .stat-bubble .sub{font-family:'Playfair Display',serif;font-size:.85rem;color:#a8a29e;margin-top:.25rem;font-style:italic;}

  /* ── chart containers ── */
  .chart-wrap{position:relative;height:280px;width:100%;}
  .chart-card{padding:1.5rem 1.5rem 1rem;}

  /* ── table ── */
  .dino-table{width:100%;border-collapse:separate;border-spacing:0;font-size:.9rem;}
  .dino-table thead tr{background:var(--green-dark);}
  .dino-table thead th{color:#fefce8;font-family:'Playfair Display',serif;font-weight:700;padding:.7rem 1rem;text-align:left;letter-spacing:.04em;font-size:.85rem;}
  .dino-table thead th:first-child{border-radius:.4rem 0 0 .4rem;}
  .dino-table thead th:last-child{border-radius:0 .4rem .4rem 0;}
  .dino-table tbody tr{transition:background .15s;}
  .dino-table tbody tr:hover{background:var(--green-light);}
  .dino-table tbody tr:nth-child(even){background:#fafaf9;}
  .dino-table tbody td{font-family:'Playfair Display',serif;font-size:.95rem;padding:.7rem 1rem;border-bottom:1px solid var(--rule);}

  /* ── win-rate bar ── */
  .winbar{height:8px;border-radius:9999px;background:#e7e5e4;overflow:hidden;min-width:80px;}
  .winbar-fill{height:100%;border-radius:9999px;transition:width 1s cubic-bezier(.4,0,.2,1);
    background:linear-gradient(90deg,var(--green),var(--yellow));}

  /* ── signal badge ── */
  .signal{
    display:inline-flex;align-items:center;gap:.4rem;
    font-family:'Playfair Display',serif;font-style:italic;
    font-size:.9rem;font-weight:500;padding:.45rem 1rem;
    border-radius:.4rem;margin-top:1rem;border:1px solid transparent;
  }
  .signal.ok{background:#f0fdf4;color:var(--green-dark);border-color:#86efac;}
  .signal.warn{background:#fef3c7;color:var(--yellow-dark);border-color:#fcd34d;}
  .signal.alert{background:#fee2e2;color:var(--red);border-color:#fca5a5;}

  /* ── dino type badge ── */
  .type-badge{
    display:inline-block;font-size:.7rem;font-weight:700;
    padding:.22rem .6rem;border-radius:.3rem;letter-spacing:.06em;
  }
  .type-Agile{background:#f0fdf4;color:var(--green-dark);border:1px solid #86efac;}
  .type-Tank{background:#eff6ff;color:#1e3a8a;border:1px solid #93c5fd;}
  .type-Speedy{background:#fef3c7;color:var(--yellow-dark);border:1px solid #fcd34d;}
  .type-Balanced{background:#f5f3ff;color:#6b21a8;border:1px solid #c4b5fd;}
  .type-default{background:#f5f5f4;color:#44403c;border:1px solid #d6d3d1;}

  /* ── habitat matrix ── */
  .matrix-grid{display:grid;gap:.5rem;}
  .matrix-row{display:flex;gap:.5rem;align-items:center;flex-wrap:wrap;}
  .matrix-cell{
    font-family:'Playfair Display',serif;font-size:.88rem;font-weight:500;
    padding:.4rem .8rem;border-radius:.35rem;border:1px solid;
    transition:transform .15s;
  }
  .matrix-cell strong{font-weight:700;}
  .matrix-cell:hover{transform:translateY(-1px);}

  /* ── traffic KPIs ── */
  .traffic-header{display:flex;gap:1rem;flex-wrap:wrap;margin-bottom:1.25rem;}
  .traffic-kpi{
    background:var(--card);color:var(--green-dark);
    border:1px solid var(--rule);border-left:4px solid var(--green-dark);
    border-radius:.4rem;padding:1rem 1.25rem;flex:1;min-width:140px;
  }
  .traffic-kpi .kv{font-family:'Playfair Display',serif;font-weight:700;font-size:1.9rem;line-height:1.1;}
  .traffic-kpi .kl{font-family:'Playfair Display',serif;font-size:.78rem;font-weight:500;color:var(--muted);letter-spacing:.22em;text-transform:uppercase;margin-top:.35rem;}

  /* ── footer ── */
  footer{
    position:relative;z-index:1;text-align:center;padding:2rem 1rem;
    font-family:'Playfair Display',serif;font-style:italic;
    font-size:.85rem;color:var(--muted);
    border-top:1px solid var(--rule);margin-top:2rem;
  }
</style>
</head>
<body>

<!-- ══ HEADER ══ -->
<header>
  <div class="header-eyebrow">Field Journal</div>
  <div class="header-title">DinoQuest Insights</div>
  <div class="header-rule"></div>
  <div class="header-sub">Game Analytics Report</div>
  <div class="header-badge">Updated /* REPORT_TIME */</div>
</header>

<main>

  <!-- ══ SECTION 1: TRAFFIC ══ -->
  <div class="section-title">🌐 Traffic Overview <span class="pill">LAST 24 HOURS</span></div>

  <div class="traffic-header">
    <div class="traffic-kpi"><div class="kv">/* TOTAL_REQUESTS */</div><div class="kl">Total Requests</div></div>
    <div class="traffic-kpi" style="border-left-color:var(--red);color:var(--red);">
      <div class="kv">/* TOTAL_ERRORS */</div><div class="kl">Server Errors</div>
    </div>
    <div class="traffic-kpi" style="border-left-color:var(--yellow-dark);color:var(--yellow-dark);">
      <div class="kv">/* AVG_LATENCY */ms</div><div class="kl">Avg Latency</div>
    </div>
  </div>

  <div class="card chart-card">
    <div class="chart-wrap"><canvas id="trafficChart"></canvas></div>
  </div>

  <!-- ══ SECTION 2: WIN RATE ══ -->
  <div class="section-title">🏆 Win Rate by Dino Type <span class="pill">DIFFICULTY</span></div>
  <div class="cards-grid" id="winRateCards">
    <!-- generated by JS -->
  </div>
  <div class="card chart-card">
    <div class="chart-wrap"><canvas id="winRateChart"></canvas></div>
  </div>

  <!-- ══ SECTION 3: TREATS ══ -->
  <div class="section-title">🍖🌿 Treats per Outcome <span class="pill">PROGRESSION</span></div>
  <div class="cards-grid">
    <div class="card stat-bubble">
      <div class="value">/* WIN_AVG_COINS */</div>
      <div class="label">Avg Treats on WIN</div>
      <div class="sub">/* WIN_GAMES */ games</div>
    </div>
    <div class="card stat-bubble">
      <div class="value" style="color:var(--red);">/* LOSS_AVG_COINS */</div>
      <div class="label">Avg Treats on LOSS</div>
      <div class="sub">/* LOSS_GAMES */ games</div>
    </div>
    <div class="card chart-card">
      <div class="chart-wrap" style="height:200px;"><canvas id="coinsChart"></canvas></div>
    </div>
  </div>
  <div id="coinSignal"></div>

  <!-- ══ SECTION 4: REUSE ══ -->
  <div class="section-title">🔁 Top Reused Dinos <span class="pill">PERSONALIZATION</span></div>
  <div class="card" style="overflow-x:auto;">
    <table class="dino-table" id="reuseTable">
      <thead><tr>
        <th>Dino Name</th><th>Type</th>
        <th>Total Starts</th><th>Reuse</th><th>Fresh</th><th>Reuse Rate</th>
      </tr></thead>
      <tbody id="reuseBody"></tbody>
    </table>
  </div>

  <!-- ══ SECTION 5: HABITAT MATRIX ══ -->
  <div class="section-title">🗺️ Habitat × Diet → Type <span class="pill">ENGINE</span></div>
  <div class="card" id="matrixCard">
    <div id="habitatMatrix"></div>
  </div>

</main>

<footer>DinoQuest Analytics &middot; Generated by Claude Code &middot; /* REPORT_TIME */</footer>

<script>
/* ══════════════════════════════════════════════
   DATA — replace these with real query results
══════════════════════════════════════════════ */

const TRAFFIC = /* TRAFFIC_JSON */;
// e.g. [{ minute:"2024-01-01 12:00", request_count:42, server_errors:1, client_errors:2, avg_latency_ms:145 }, ...]

const WIN_RATE = /* WIN_RATE_JSON */;
// e.g. [{ dino_type:"Agile", total_games:50, wins:38, win_rate_pct:76.0, avg_score:320, avg_speed:1.8 }, ...]

const COINS = /* COINS_JSON */;
// e.g. [{ won:true, games:38, avg_coins:4.2, min_coins:1, max_coins:9 }, { won:false, games:12, avg_coins:1.1, min_coins:0, max_coins:3 }]

const REUSE = /* REUSE_JSON */;
// e.g. [{ dino_name:"Rex", dino_type:"Tank", total_starts:15, reuse_starts:11, fresh_starts:4 }, ...]

const HABITAT = /* HABITAT_JSON */;
// e.g. [{ habitat:"forest", diet:"herbivore", dino_type:"Agile", occurrences:8 }, ...]

/* ══════════════════════════════════════════════
   CHART DEFAULTS
══════════════════════════════════════════════ */
Chart.defaults.font.family = "'Playfair Display', serif";
Chart.defaults.font.weight = '500';
Chart.defaults.font.size = 13;
Chart.defaults.color = '#292524';
Chart.defaults.borderColor = 'rgba(0,0,0,.06)';

/* ══════════════════════════════════════════════
   TRAFFIC CHART
══════════════════════════════════════════════ */
if(TRAFFIC && TRAFFIC.length){
  const labels = TRAFFIC.map(r=>r.minute.slice(11,16)).reverse();
  const reqs   = TRAFFIC.map(r=>r.request_count).reverse();
  const errs   = TRAFFIC.map(r=>r.server_errors).reverse();
  const lat    = TRAFFIC.map(r=>r.avg_latency_ms).reverse();
  new Chart(document.getElementById('trafficChart'),{
    type:'bar',
    data:{
      labels,
      datasets:[
        { label:'Requests', data:reqs, backgroundColor:'rgba(21,128,61,.65)', borderColor:'#15803d', borderWidth:1.5, borderRadius:3, yAxisID:'y' },
        { label:'Server Errors', data:errs, backgroundColor:'rgba(185,28,28,.65)', borderColor:'#b91c1c', borderWidth:1.5, borderRadius:3, yAxisID:'y' },
        { label:'Latency (ms)', data:lat, type:'line', borderColor:'#b45309', backgroundColor:'rgba(180,83,9,.10)', borderWidth:2, pointRadius:3, pointHoverRadius:6, tension:.35, yAxisID:'y1' }
      ]
    },
    options:{
      responsive:true,maintainAspectRatio:false,
      interaction:{mode:'index',intersect:false},
      plugins:{legend:{labels:{color:'#292524',padding:16}},tooltip:{backgroundColor:'#14532d',titleColor:'#fde68a',bodyColor:'#fefce8',cornerRadius:6,padding:10}},
      scales:{
        y:{beginAtZero:true,grid:{color:'rgba(0,0,0,.05)'},ticks:{color:'#78716c'}},
        y1:{position:'right',beginAtZero:true,grid:{display:false},ticks:{color:'#b45309'}}
      },
      animation:{duration:1000,easing:'easeOutQuart'}
    }
  });
}

/* ══════════════════════════════════════════════
   WIN RATE CARDS + CHART
══════════════════════════════════════════════ */
const TYPE_COLORS = { Agile:'#15803d', Tank:'#1e40af', Speedy:'#d97706', Balanced:'#7c3aed' };
function typeColor(t){ return TYPE_COLORS[t] || '#78716c'; }

if(WIN_RATE && WIN_RATE.length){
  const cardsEl = document.getElementById('winRateCards');
  WIN_RATE.forEach((r,i)=>{
    const pct = r.win_rate_pct||0;
    const hot = pct > 70;
    const card = document.createElement('div');
    card.className = 'card stat-bubble';
    card.style.borderColor = typeColor(r.dino_type);
    card.style.animationDelay = (i*0.12)+'s';
    card.innerHTML = `
      <div style="font-size:2rem;margin-bottom:.4rem;opacity:.7;">${{Agile:'⚡',Tank:'🛡️',Speedy:'💨',Balanced:'⚖️'}[r.dino_type]||'◆'}</div>
      <div class="value" style="font-size:2.4rem;">${pct}%</div>
      <div class="label">${r.dino_type} Win Rate</div>
      <div class="sub">${r.total_games} games · avg score ${r.avg_score}</div>
      ${hot?'<div class="signal alert">⚠ Needs rebalance</div>':'<div class="signal ok">✓ Balanced</div>'}
    `;
    cardsEl.appendChild(card);
  });

  new Chart(document.getElementById('winRateChart'),{
    type:'radar',
    data:{
      labels:WIN_RATE.map(r=>r.dino_type),
      datasets:[
        { label:'Win Rate %', data:WIN_RATE.map(r=>r.win_rate_pct), backgroundColor:'rgba(21,128,61,.20)', borderColor:'#15803d', borderWidth:2, pointBackgroundColor:WIN_RATE.map(r=>typeColor(r.dino_type)), pointRadius:5 },
        { label:'Avg Score/10', data:WIN_RATE.map(r=>(r.avg_score/10).toFixed(1)), backgroundColor:'rgba(180,83,9,.15)', borderColor:'#b45309', borderWidth:1.5, pointBackgroundColor:'#b45309', pointRadius:4 }
      ]
    },
    options:{
      responsive:true,maintainAspectRatio:false,
      plugins:{legend:{labels:{color:'#292524'}},tooltip:{backgroundColor:'#14532d',titleColor:'#fde68a',bodyColor:'#fefce8',cornerRadius:6}},
      scales:{r:{beginAtZero:true,max:100,grid:{color:'rgba(0,0,0,.06)'},pointLabels:{color:'#292524',font:{size:14,weight:'700'}},ticks:{backdropColor:'transparent',color:'#a8a29e'}}},
      animation:{duration:1100,easing:'easeOutQuart'}
    }
  });
}

/* ══════════════════════════════════════════════
   COINS CHART + SIGNAL
══════════════════════════════════════════════ */
if(COINS && COINS.length){
  const winRow  = COINS.find(r=>r.won===true||r.won==='true')||{};
  const lossRow = COINS.find(r=>r.won===false||r.won==='false')||{};
  new Chart(document.getElementById('coinsChart'),{
    type:'doughnut',
    data:{
      labels:['Win Avg Treats','Loss Avg Treats'],
      datasets:[{ data:[winRow.avg_coins||0, lossRow.avg_coins||0], backgroundColor:['#15803d','#b91c1c'], borderColor:['#14532d','#7f1d1d'], borderWidth:2, hoverOffset:8 }]
    },
    options:{
      responsive:true,maintainAspectRatio:false,cutout:'65%',
      plugins:{legend:{labels:{color:'#292524',padding:12}},tooltip:{backgroundColor:'#14532d',titleColor:'#fde68a',bodyColor:'#fefce8',cornerRadius:6}},
      animation:{animateRotate:true,duration:1000,easing:'easeOutQuart'}
    }
  });
  const winCoins = winRow.avg_coins||0;
  const sig = document.getElementById('coinSignal');
  if(winCoins<=3){
    sig.innerHTML='<div class="signal warn" style="font-size:.9rem;padding:.55rem 1rem;">⚠ Win avg treats ≤ 3 — consider a streak bonus to boost retention</div>';
  } else {
    sig.innerHTML='<div class="signal ok" style="font-size:.9rem;padding:.55rem 1rem;">✓ Treat yield looks healthy — players have progression incentive</div>';
  }
}

/* ══════════════════════════════════════════════
   REUSE TABLE
══════════════════════════════════════════════ */
if(REUSE && REUSE.length){
  const tbody = document.getElementById('reuseBody');
  REUSE.forEach((r,i)=>{
    const reuseRate = r.total_starts>0 ? Math.round(r.reuse_starts/r.total_starts*100) : 0;
    const typeCls = `type-${r.dino_type}`;
    const row = document.createElement('tr');
    row.innerHTML = `
      <td><strong>${r.dino_name}</strong></td>
      <td><span class="type-badge ${typeCls}">${r.dino_type||'—'}</span></td>
      <td>${r.total_starts}</td>
      <td><strong style="color:var(--green-dark);">${r.reuse_starts}</strong></td>
      <td>${r.fresh_starts}</td>
      <td>
        <div style="display:flex;align-items:center;gap:.5rem;">
          <div class="winbar"><div class="winbar-fill" style="width:0%" data-pct="${reuseRate}"></div></div>
          <span style="font-family:'Playfair Display',serif;font-weight:700;color:var(--ink);">${reuseRate}%</span>
        </div>
      </td>
    `;
    tbody.appendChild(row);
  });
  requestAnimationFrame(()=>{
    document.querySelectorAll('.winbar-fill').forEach(el=>{
      setTimeout(()=>{ el.style.width=el.dataset.pct+'%'; }, 200);
    });
  });
}

/* ══════════════════════════════════════════════
   HABITAT MATRIX
══════════════════════════════════════════════ */
if(HABITAT && HABITAT.length){
  const HABITAT_COLORS={forest:'#14532d',desert:'#b45309',swamp:'#065f46',beach:'#0e7490',tundra:'#475569',volcano:'#b91c1c',ocean:'#1e40af'};
  const DIET_COLORS={herbivore:'#15803d',carnivore:'#b91c1c',omnivore:'#b45309'};
  const matrix = document.getElementById('habitatMatrix');
  const grouped={};
  HABITAT.forEach(r=>{
    const key=`${r.habitat}|${r.diet}`;
    if(!grouped[key]) grouped[key]={habitat:r.habitat,diet:r.diet,types:[]};
    grouped[key].types.push({type:r.dino_type,n:r.occurrences});
  });
  const habitats=[...new Set(HABITAT.map(r=>r.habitat))];
  const diets=[...new Set(HABITAT.map(r=>r.diet))];
  const grid = document.createElement('div');
  grid.style.cssText='display:grid;gap:.75rem;';
  habitats.forEach(h=>{
    const row=document.createElement('div');
    row.className='matrix-row';
    const hLabel=document.createElement('div');
    hLabel.style.cssText=`font-family:'Playfair Display',serif;font-weight:700;font-size:1rem;min-width:90px;color:${HABITAT_COLORS[h]||'var(--green-dark)'};`;
    hLabel.textContent='🗺️ '+h;
    row.appendChild(hLabel);
    diets.forEach(d=>{
      const key=`${h}|${d}`;
      const entry=grouped[key];
      if(!entry) return;
      const allSameType = new Set(entry.types.map(t=>t.type)).size===1;
      const cell=document.createElement('div');
      cell.className='matrix-cell';
      const bg=DIET_COLORS[d]||'#6b7280';
      cell.style.cssText=`background:${bg}18;color:${bg};border-color:${bg};`;
      const typeSummary=entry.types.map(t=>`${t.type}(${t.n})`).join(', ');
      cell.innerHTML=`<span>${d}</span> → <strong>${typeSummary}</strong>${allSameType?' <span style="color:var(--yellow-dark);font-weight:700;">★</span>':''}`;
      cell.title=allSameType?'Deterministic — good candidate for type preview UI':'Multiple outcomes';
      row.appendChild(cell);
    });
    grid.appendChild(row);
  });
  matrix.appendChild(grid);
  const deterministicCombos=Object.values(grouped).filter(g=>new Set(g.types.map(t=>t.type)).size===1);
  if(deterministicCombos.length>0){
    const sig=document.createElement('div');
    sig.className='signal ok';
    sig.style.marginTop='1rem';
    sig.innerHTML=`★ ${deterministicCombos.length} habitat + diet combo(s) are 100% deterministic — show type preview before confirm`;
    matrix.appendChild(sig);
  }
}

/* ══════════════════════════════════════════════
   ANIMATED NUMBER COUNTERS
══════════════════════════════════════════════ */
function animateCount(el, target, suffix=''){
  const duration=1200, start=performance.now(), from=0;
  function step(now){
    const t=Math.min((now-start)/duration,1);
    const ease=1-Math.pow(1-t,3);
    el.textContent=Math.round(from+(target-from)*ease)+suffix;
    if(t<1) requestAnimationFrame(step);
  }
  requestAnimationFrame(step);
}
document.querySelectorAll('.value[data-count]').forEach(el=>{
  animateCount(el, parseFloat(el.dataset.count), el.dataset.suffix||'');
});
</script>
</body>
</html>

---

Data injection rules

When writing the file, replace each placeholder with a JS literal constructed from real query data:

Placeholder	Replace with
/* REPORT_TIME */	ISO-8601 string from the current timestamp
/* TOTAL_REQUESTS */	sum of request_count across all traffic rows
/* TOTAL_ERRORS */	sum of server_errors across all traffic rows
/* AVG_LATENCY */	average of avg_latency_ms across all traffic rows, rounded
/* TRAFFIC_JSON */	JSON array from Step 2.2 results
/* WIN_RATE_JSON */	JSON array from Step 3.1 results
/* COINS_JSON */	JSON array from Step 3.2 results
/* REUSE_JSON */	JSON array from Step 3.3 results
/* HABITAT_JSON */	JSON array from Step 3.4 results
/* WIN_AVG_COINS */	avg_coins where won=true, 1 decimal
/* WIN_GAMES */	games count where won=true
/* LOSS_AVG_COINS */	avg_coins where won=false, 1 decimal
/* LOSS_GAMES */	games count where won=false

If any query returned no data, substitute an empty array [] for that dataset — the JS gracefully skips empty sections.

---

Edge Cases

• Dataset/table not found after sink creation: Logs take 1–5 minutes to appear. Retry Phase 2 after waiting.
• BigQuery execute_sql returns empty results: No traffic in the last hour, or the sink was just created. Try extending the INTERVAL to INTERVAL 24 HOUR.
• Sink already exists but BigQuery has no data: Verify the writerIdentity has bigquery.dataEditor on the dataset by checking IAM (run scripts/check_sink.sh for diagnostics).
• MCP server disabled: Enable google-cloud-logging and google-bigquery in mcp_config.json and restart the agent before retrying.
• Phase 3 Unrecognized name: jsonPayload: There is no jsonPayload column. The backend uses print(json.dumps({...})), so events are plain JSON strings in textPayload. Use JSON_VALUE(textPayload, '$.field') for all field access.
• Phase 3 boolean fields: won and is_reuse are JSON string values, not native BOOL. Compare with = 'true' / = 'false', not = TRUE.
• Phase 3 DINO_CREATED returns empty: Events only fire on new dino generation. Always use the 7-day wildcard query from Step 3.4 to aggregate across multiple daily tables.
• Phase 3 numeric cast errors: Extract with JSON_VALUE first, then cast: SAFE_CAST(JSON_VALUE(textPayload, '$.score') AS FLOAT64).