Skip to main content

​
What this is for

The Logs page is the operational view of traffic flowing through your gateway. It captures four complementary perspectives, organized as tabs:
  • Requests — flat, filterable table of every HTTP request the gateway served.
  • MCP — every tool call made through an MCP server attached to the gateway.
  • Guardrails — every guardrail rule trigger (input or output, blocked or logged).
  • Traces — OpenTelemetry timelines showing how each request flowed through guardrails, auth, routing, cache, and the upstream provider.
Request, MCP, guardrail, and trace logs are recorded on the gateway. Viewing them requires the dashboard to be connected to that gateway — see Activation. Administrative audit logs live on the platform — see Audit Log.

​
Requests

The default tab. A flat table of every HTTP request, sortable on any column.

​
Columns

Timestamp, method, path, status, duration, provider, model, API key name, tokens (prompt + completion), cost, client IP, and trace ID.

​
Filters

Requests log table

Logs → Requests

Click a row’s trace ID to jump straight to the matching trace.

​
MCP

Every call to a tool exposed by an MCP server registered with the gateway is logged here. Filters mirror Requests: time range, gateway, and a free-text search that matches tool name, MCP server, or API key name.
MCP logs

Logs → MCP

If a call shows denied, check the API key’s MCP scope on API Keys — the key’s allow-list is the usual cause.

​
Guardrails

Every guardrail rule trigger — both rule-based (Guardrails) and SLM-based (SLM Guardrails) — is recorded here. Each row shows the rule name, category, action taken (block, warn, redact, log), the matched content (or its hash for redacted entries), and a link back to the originating request. Use this tab to audit policy hits, tune thresholds before promoting an SLM model to a blocking action, and confirm a rule is matching what you expect.
Guardrails logs

Logs → Guardrails

​
Traces

A trace is a timeline of the steps a request went through inside the gateway. Each step has an icon: Click a step to see its latency, tokens consumed, cost contribution, and any error. The entire trace is time-aligned, so a slow guardrail or a stuck provider call stands out immediately.
Trace detail

Logs → Traces

​
How to configure

There is no configuration on this page itself — Logs is read-only. To change what gets captured:
1

Add or rotate API keys

Configuration → API Keys determines which key labels show up in the API Key filter.
2

Tune guardrails

Configuration → Security controls what shows up in the Guardrails tab.
3

Adjust retention

Log retention is set per gateway in the gateway environment. See Environment.

​
Exports

From any tab, click Export to download the current filter as CSV or JSON. Exports are assembled locally on the gateway; the dashboard only forwards the query.