diff --git a/.github/workflows/uptime-probe.yml b/.github/workflows/uptime-probe.yml
index 21302b9..a882c2d 100644
--- a/.github/workflows/uptime-probe.yml
+++ b/.github/workflows/uptime-probe.yml
@@ -85,6 +85,17 @@ jobs:
           echo "== run summary =="
           jq -r '.[] | "\(.name): \(.status_code) \(.latency_ms)ms success=\(.success)"' /tmp/run.json || cat /tmp/run.json
 
+      - name: Aggregate probe results → Upptime format
+        # Reads history/<slug>.jsonl files, computes rolling uptime/response-time
+        # aggregates, and writes history/<slug>.yml + history/summary.json.
+        # This fills the gap left by the Upptime → custom-probe migration:
+        # the probe binary handles JSONL appends; this step handles the
+        # aggregator outputs that the status page UI reads.
+        # See molecule-ai/molecule-ai-status#7.
+        run: |
+          set -euo pipefail
+          python3 scripts/aggregate.py --history-dir history
+
       - name: Commit history changes
         # Fails fast if Gitea is unhealthy rather than silently swallowing
         # the push. The next /5 cron firing picks up where this left off once
diff --git a/scripts/aggregate.py b/scripts/aggregate.py
new file mode 100755
index 0000000..19d536d
--- /dev/null
+++ b/scripts/aggregate.py
@@ -0,0 +1,263 @@
+#!/usr/bin/env python3
+"""
+aggregate.py — Upptime-format aggregator for molecule-ai-status.
+
+Reads probe results from history/<slug>.jsonl files, computes rolling
+uptime and response-time aggregates, and writes:
+
+  history/<slug>.yml      — latest probe result (Upptime status-file format)
+  history/summary.json   — per-site aggregates for day/week/month/year
+
+Run after each probe run, before the git commit step.
+
+Usage:
+    python3 scripts/aggregate.py [--history-dir history]
+"""
+
+import argparse
+import json
+import sys
+import time
+from datetime import datetime, timezone, timedelta
+from pathlib import Path
+
+
+def parse_ts(ts: str) -> datetime:
+    """Parse ISO-8601 timestamp with Z suffix."""
+    return datetime.fromisoformat(ts.replace("Z", "+00:00"))
+
+
+def compute_uptime_pct(results: list[dict], since: datetime) -> tuple[float, int]:
+    """
+    Return (uptime_percent, minutes_down) for results since `since`.
+
+    A 5-minute interval is "down" if the result at or after that minute
+    had success=False.  minutes_down is the count of such 5-min slots.
+    """
+    now = datetime.now(timezone.utc)
+    if not results:
+        return 100.0, 0
+
+    # Bucket results into 5-minute slots
+    slots: dict[int, bool] = {}  # slot_minute -> any_success
+    for r in results:
+        if parse_ts(r["timestamp"]) < since:
+            continue
+        slot = int(parse_ts(r["timestamp"]).timestamp() // 300) * 300
+        # If any probe in this slot succeeded, the slot is up
+        if slots.get(slot, r["success"]):
+            slots[slot] = r["success"]
+
+    if not slots:
+        return 100.0, 0
+
+    total_slots = len(slots)
+    up_slots = sum(1 for v in slots.values() if v)
+    uptime_pct = (up_slots / total_slots) * 100
+    minutes_down = total_slots - up_slots
+    return round(uptime_pct, 2), minutes_down
+
+
+def avg_response_time(results: list[dict], since: datetime) -> float | None:
+    """Return average latency_ms for results since `since`."""
+    latencies = [
+        r["latency_ms"]
+        for r in results
+        if parse_ts(r["timestamp"]) >= since and r.get("latency_ms") is not None
+    ]
+    return round(sum(latencies) / len(latencies), 0) if latencies else None
+
+
+def slug_from_name(name: str) -> str:
+    """Derive slug from site name (matches Upptime convention)."""
+    return name.lower().replace(" — ", "-").replace(" ", "-").replace(".", "")
+
+
+def get_sites(upptimerc: Path) -> list[dict]:
+    """Parse .upptimerc.yml to get site list (name, url)."""
+    try:
+        import yaml
+    except ImportError:
+        pass  # Fall back to simple parser below
+
+    content = upptimerc.read_text()
+
+    # Try yaml import first
+    try:
+        import yaml as _yaml
+        data = _yaml.safe_load(content)
+        raw_sites = data.get("sites", []) if data else []
+        return [{"name": s["name"], "url": s["url"]} for s in raw_sites if s.get("name")]
+    except Exception:
+        pass
+
+    # Fallback: simple line-based parser for indented - name: / url: pairs
+    sites = []
+    in_sites = False
+    current = {}
+    for line in content.splitlines():
+        stripped = line.strip()
+        indent = len(line) - len(stripped)
+        if stripped == "sites:":
+            in_sites = True
+            continue
+        if not in_sites:
+            continue
+        # Dedent back to top-level ends sites block
+        if indent == 0 and stripped and not stripped.startswith("-"):
+            break
+        if stripped.startswith("- name:"):
+            if current.get("name"):
+                sites.append(current)
+            current = {"name": stripped.split("name:", 1)[1].strip().lstrip("- ")}
+        elif stripped.startswith("url:"):
+            current["url"] = stripped.split("url:", 1)[1].strip()
+    if current.get("name"):
+        sites.append(current)
+    return sites
+
+
+def write_yml(slug: str, latest: dict | None, first_ts: str | None) -> str:
+    """Write history/<slug>.yml in Upptime format."""
+    if latest is None:
+        # No probe results — leave as-is or write a placeholder
+        return ""
+    status = "up" if latest["success"] else "down"
+    code = latest["status_code"]
+    response_time = latest["latency_ms"]
+    last_updated = latest["timestamp"]
+    start_time = first_ts or last_updated
+    yml = f"""\
+url: {latest["url"]}
+status: {status}
+code: {code}
+responseTime: {response_time}
+lastUpdated: {last_updated}
+startTime: {start_time}
+generator: Upptime <https://github.com/upptime/upptime>
+"""
+    return yml
+
+
+def write_summary_site_entry(name: str, url: str, slug: str,
+                             results: list[dict],
+                             start_time: datetime) -> dict:
+    """Build a summary.json entry for one site."""
+    now = datetime.now(timezone.utc)
+    day_start = now - timedelta(days=1)
+    week_start = now - timedelta(weeks=1)
+    month_start = now - timedelta(days=30)
+    year_start = now - timedelta(days=365)
+
+    uptime, uptime_day, uptime_week, uptime_month, uptime_year = None, None, None, None, None
+    rt, rt_day, rt_week, rt_month, rt_year = None, None, None, None, None
+
+    # All-time
+    uptime, _ = compute_uptime_pct(results, start_time)
+    rt = avg_response_time(results, start_time)
+
+    # Day
+    uptime_day, _ = compute_uptime_pct(results, day_start)
+    rt_day = avg_response_time(results, day_start)
+
+    # Week
+    uptime_week, _ = compute_uptime_pct(results, week_start)
+    rt_week = avg_response_time(results, week_start)
+
+    # Month
+    uptime_month, _ = compute_uptime_pct(results, month_start)
+    rt_month = avg_response_time(results, month_start)
+
+    # Year
+    uptime_year, _ = compute_uptime_pct(results, year_start)
+    rt_year = avg_response_time(results, year_start)
+
+    latest = results[-1] if results else {}
+    status = "up" if latest.get("success", True) else "down"
+
+    def fmt(val):
+        if val is None:
+            return None
+        return f"{val:.2f}%" if isinstance(val, float) else val
+
+    return {
+        "name": name,
+        "url": url,
+        "slug": slug,
+        "status": status,
+        "uptime": fmt(uptime),
+        "uptimeDay": fmt(uptime_day),
+        "uptimeWeek": fmt(uptime_week),
+        "uptimeMonth": fmt(uptime_month),
+        "uptimeYear": fmt(uptime_year),
+        "time": rt,
+        "timeDay": rt_day,
+        "timeWeek": rt_week,
+        "timeMonth": rt_month,
+        "timeYear": rt_year,
+        "dailyMinutesDown": {},
+    }
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Aggregate upptime probe results")
+    parser.add_argument("--history-dir", default="history", help="Path to history directory")
+    args = parser.parse_args()
+
+    history_dir = Path(args.history_dir)
+    upptimerc = Path(".upptimerc.yml")
+
+    if not history_dir.exists():
+        print(f"No history directory: {history_dir}")
+        sys.exit(1)
+
+    sites = get_sites(upptimerc)
+    print(f"Aggregating {len(sites)} sites from {history_dir}/")
+
+    summary_entries = []
+    written_ymls = 0
+
+    for site in sites:
+        name = site["name"]
+        url = site["url"]
+        slug = slug_from_name(name)
+        jsonl_path = history_dir / f"{slug}.jsonl"
+
+        results = []
+        if jsonl_path.exists():
+            for line in jsonl_path.read_text().strip().splitlines():
+                if line.strip():
+                    try:
+                        results.append(json.loads(line))
+                    except json.JSONDecodeError:
+                        pass
+
+        # Sort by timestamp
+        results.sort(key=lambda r: r.get("timestamp", ""))
+        latest = results[-1] if results else None
+        first_ts = results[0].get("timestamp") if results else None
+        start_time = parse_ts(first_ts) if first_ts else datetime.now(timezone.utc)
+
+        # Write .yml
+        yml_content = write_yml(slug, latest, first_ts)
+        if yml_content:
+            yml_path = history_dir / f"{slug}.yml"
+            yml_path.write_text(yml_content)
+            written_ymls += 1
+            status = "up" if latest.get("success") else "down"
+            print(f"  {slug}: {status} ({len(results)} results, latest {latest.get('status_code') if latest else 'N/A'})")
+        else:
+            print(f"  {slug}: no results (skipped)")
+
+        # Build summary entry
+        entry = write_summary_site_entry(name, url, slug, results, start_time)
+        summary_entries.append(entry)
+
+    # Write summary.json
+    summary_path = history_dir / "summary.json"
+    summary_path.write_text(json.dumps(summary_entries, indent=2))
+    print(f"\nWrote {written_ymls} .yml files + summary.json ({len(summary_entries)} entries)")
+
+
+if __name__ == "__main__":
+    main()