TRANSPARENCY · LIVE
Status
Public telemetry for aethelforge.ai and the gate. If it's not on this page, we're not measuring it. If we're not measuring it, you shouldn't trust it.
AEO · atomic answers
7pages
reporting
AEO · schema coverage
100%
reporting
CWV · LCP (p75)
—s
pending wire
CWV · INP (p75)
—ms
pending wire
CWV · CLS (p75)
—
pending wire
Crawl · allow-listed bots
11
reporting
Gate · handshakes last 24h
—
pending wire
Gate · p95 latency
—ms
pending wire
What's here
This page is the honest answer to “does the gate actually work?” Public telemetry, not marketing bullets. Numbers marked pending wireare metrics we plan to surface here but haven't yet — the infrastructure is staged, the endpoints aren't connected. As each lands, the value appears. No backfill, no retroactive smoothing.
What's coming
- Core Web Vitals (LCP / INP / CLS) wired from the real-user monitoring beacon. Target: p75 LCP under 2.0s, INP under 200ms, CLS under 0.1.
- Gate telemetry: handshake count, p95/p99 latency, success rate, protocol mix across MCP / ACP / AP2 / x402. Redacted of tenant identifiers.
- AEO audit: per-page atomic-answer word count, schema emission check, llms.txt freshness, robots.txt crawler policy.
- Infrastructure: CDN hit rate, Lambda cold-start histogram, CloudFront error rate, SES deliverability.
Why this page exists
Because trust requires something to check. A competent gate publishes its own numbers. If we ever stop doing that, stop trusting us.