Telemetry Bootstrap + Real Metrics Baseline Capture

First live telemetry run for the AI Execution Lab. The telemetry infrastructure has been built for several sessions — this is the first time real data was captured and stored.

What Was Run

Uptime check

node scripts/ingest-uptime.mjs

Result: 5/5 properties up. All HTTP 200.

Data written to data/telemetry/uptime.json. First verified uptime baseline at 2026-05-19T04:58:07Z.

The project.asquaresolution.com latency stands out at 1047ms — 4× slower than the Lab. This subdomain has no active product and receives no optimization. Not critical, but it's the first latency anomaly in the baseline.

Sitemap full check

node scripts/ingest-sitemap.mjs --full

Result: 313/313 URLs passing. 0 failures. Sitemap HTTP 200.

Verified every URL in the lab's sitemap individually with HEAD requests. Full mode takes ~60 seconds. Result: zero 404s, zero redirect chains, zero unreachable content.

313 published URLs at baseline. Every future sitemap run will show whether this grew or if any pages became unreachable.

Data written to data/telemetry/sitemap-health.json.

What Was Blocked

Lighthouse — PSI API quota exceeded

node scripts/ingest-lighthouse.mjs
# [SKIP] PSI API 429: Quota exceeded

The PageSpeed Insights API has a daily quota for unauthenticated requests. The quota was already exhausted when the script ran (likely from earlier testing during the build session).

Fix: Set PAGESPEED_API_KEY=<your_key> in .env.local. Free Google API keys have a quota of 25,000 requests/day — more than sufficient. Get key from Google Cloud Console → APIs & Services → Credentials.

Without the API key, the script is limited to ~1-2 requests/day on the anonymous quota. This is a one-time setup task.

GSC — no CSV export yet

GSC data requires manually exporting from Google Search Console and running the ingestion script against the CSV file. This can't be automated without OAuth (which is out of scope for this platform's architecture).

Next step:

1. Go to Google Search Console → Performance → Search Results
2. Set date range: last 28 days
3. Click Export → Download CSV (repeat for Queries tab and Pages tab)
4. Run: node scripts/ingest-gsc.mjs path/to/queries.csv --type queries
5. Run: node scripts/ingest-gsc.mjs path/to/pages.csv --type pages

Until done, the GSC panel in the telemetry dashboard shows "unknown" tier.

Vercel — no env vars set

The Vercel ingestion script needs VERCEL_TOKEN and VERCEL_PROJECT_ID in .env.local. These aren't set yet.

Next step: Vercel Dashboard → Settings → Tokens → Create Token → add to .env.local → run node scripts/ingest-vercel.mjs.

Script Bug Fixed

scripts/ingest-uptime.mjs had a TypeScript type annotation (const status: 'up' | 'degraded' =) in a plain .mjs file. Node.js v25 can't parse TypeScript syntax in .mjs files. Removed the type annotation — the duplicate status computation was the dead line.

// Before (broken)
const latencyMs = Date.now() - start
const status: 'up' | 'degraded' =     // ← TypeScript syntax in .mjs
  res.ok ? 'up' : res.status >= 500 ? 'degraded' : 'up'

// After (fixed)
const latencyMs = Date.now() - start
// status computed inline in return object below

The duplicate variable was unused — status was being computed twice (once with the type annotation, once in the return object). The return object's inline computation was the correct one.

Telemetry Dashboard Now Has Real Data

The /ops/telemetry page now reads from real snapshots:

• Uptime panel: 5 real properties with HTTP status and latency
• Sitemap panel: 313/313 verified URLs
• Performance panel: still "unknown" — pending Lighthouse run with API key
• Search panel: still "unknown" — pending GSC CSV export
• Deployment panel: still "unknown" — pending Vercel env vars

Overall system health tier: warning (partial data — 2/5 subsystems have real data)

Target: all 5 subsystems active within this week.

Metrics That Now Have Baselines

These baselines have no value standing alone. Their value compounds as subsequent captures show movement. The point is not the number — it is the delta.

Next Telemetry Actions

1. PAGESPEED_API_KEY → .env.local → node scripts/ingest-lighthouse.mjs
2. GSC export → node scripts/ingest-gsc.mjs <csv>
3. VERCEL_TOKEN + VERCEL_PROJECT_ID → .env.local → node scripts/ingest-vercel.mjs
4. Commit data/telemetry/ snapshot files to repo
5. Schedule weekly uptime + monthly sitemap full-check