See every request. Hear about it when it breaks.
A traffic inspector with replay, metrics with SLO error budgets, endpoint health checks, and alert rules that post to Slack, Discord, email or your own webhook. It covers a service you expose through the ngris agent — on your laptop, a server or Kubernetes — and an app you host on ngris, from the same dashboard.
- Agent tunnels and hosted apps
- Auth headers redacted
- Fired and resolved notices
See every request, then send it again
The Traffic Inspector records each request the edge serves, for agent endpoints and hosted apps alike. The agent keeps its own copy on your machine, so you can debug locally without opening a browser tab.
replay, edited: X-Debug: 1 → 200 in 182 ms
Live tail with filters for status, method, path, host, IP, country, ASN, browser and region, or describe it in plain English and get a real filter.
AI analysis of one request: explain it, scan it for security issues, or suggest a fix. Export to JSON or CSV.
Authorization, Proxy-Authorization and Set-Cookie are stored as [redacted]; per endpoint keep metadata only, or turn capture off.
Metrics and SLOs from the traffic you actually served
Traffic Overview turns the same request records into charts, targets and a security view, so the numbers on each page agree.
Metrics
Requests, error rate, median, p95 and p99 latency, bandwidth in and out, status codes and unique visitors — live, or over a time range, with the change against the previous period.
Break it down
Per endpoint, per agent and per region. Top endpoints by requests, latency, errors and bandwidth; the most recent errors; several endpoints side by side.
SLOs and error budget
Set your account's availability target and a latency objective — a threshold and the share of requests that must beat it. The SLO tab shows the error budget left and the burn rate.
Security analytics
What your firewall and rate limits stopped, with the top source IPs, ASNs, countries, user agents, paths and rules behind it. See bot protection.
Changes to your account land in the account audit log; your own sign-ins and security changes in your activity log.
Health checks for both kinds of origin
An agent endpoint is probed from outside, the way a visitor reaches it. A backend hosted on ngris is checked from inside, by the platform that runs it.
- Endpoint health checks: ngris probes your endpoint over public HTTPS on the path and interval you choose; uptime over 24 hours, 7 and 30 days. Signed probes stay out of your traffic logs and usage.
- Down and back up: the owner gets an email when it goes down and another when it recovers, on the change of state, not every probe.
- Hosted backends: a replica that fails its health-check path leaves rotation until it recovers; a process that stops answering is restarted.
Alerts where your team already is
Write a threshold rule for the whole account or a single endpoint. ngris posts when it fires and again when it resolves.
Alert rules
request_rate,error_rate_pct,p95_latency_ms,p99_latency_ms,bandwidth_in_bps,bandwidth_out_bps- A window from 1 minute to 24 hours, and a cooldown so a flapping metric doesn't flood the channel
- Preview a rule: the metric's value right now, and whether it would fire
- Event history per rule, with the delivery status of each notice
- Or describe what to watch in plain English — “error rate over 1% for 5 minutes” — and the AI generator writes the rules
Four channels
- Email to any address
- Webhook: a JSON POST to your
https://URL, signed with HMAC-SHA256 inX-Ngris-Signaturewhen you set a secret - Slack: an incoming webhook into the channel you pick
- Discord: a channel webhook
- Send a test notification to a channel before you rely on it
Iris also watches for you in the dashboard: certificates and domains close to expiry, usage burning fast, endpoints that went offline, rising error rates or latency, and suspicious traffic.
More for apps you host on ngris
Hosted apps get the inspector, metrics and alerts above, plus the views you need once your code runs on our side.
Build logs and status
Every deploy keeps its build log. Follow a deploy's status live in the dashboard, or run ngris apps status.
Runtime logs
The stdout and stderr of your running backend, full-stack app, WordPress site or function. From the CLI: ngris apps logs --runtime, with --previous after a crash.
Web analytics
Page views, visitors, top pages, countries and devices, built from the edge's own request records: no script to add, no cookies, no per-visitor tracking.
Speed Insights
Core Web Vitals from real visitors: TTFB, FCP, LCP, CLS and INP. Switch it on and the edge adds a small collector to your pages — no cookies, no identifiers.
Frequently asked questions
Does monitoring work for agent tunnels as well as hosted apps?
Yes. The traffic inspector, metrics, SLOs, health checks and alerts cover a service you expose through the ngris agent (on your laptop, a server or Kubernetes) and an app you host on ngris, from the same dashboard. Hosted apps also get build and runtime logs, web analytics and Speed Insights.
Which alert channels are supported?
Email, Slack (an incoming webhook), Discord (a channel webhook) and your own HTTPS webhook, which can be signed with HMAC-SHA256 in the X-Ngris-Signature header. You can send a test notification to any channel before relying on it.
What can an alert rule watch?
Request rate, error rate, p95 and p99 latency, and inbound or outbound bandwidth, for the whole account or one endpoint, over a window from 1 minute to 24 hours. Preview a rule to see the metric's current value and whether it would fire; you are notified when it fires and when it resolves.
Are passwords and tokens stored in the traffic inspector?
No. Authorization, Proxy-Authorization and Set-Cookie values are stored as [redacted]. Per endpoint you can also keep only request metadata, or turn capture off entirely.
Can I debug without opening the dashboard?
Yes. While the agent runs it serves a local inspector on 127.0.0.1:9090, and ngris inspect, ngris replay and ngris export give you the same requests in your terminal, including replaying one to your local backend with edited headers.
Know before your users do.
Start free — no card needed — expose a service or deploy an app, and watch its first requests arrive in the inspector.