12:41:03hikari-apiGET /api/facets returned service buckets in 42ms

12:41:04collectoraccepted 500 events from namespace edge-staging

12:41:05paymentsretrying request after upstream timeout client=demo-shop

12:41:06auth-servicelogin failed reason=invalid_token pod=auth-7cbfcdb8f9

12:41:07mcp-servertool=query_logs query="_time:15m level:error" rows=83

12:41:08routermatched route=/orders/:id latency_ms=18 host=node-a.internal.example

12:41:09workerqueue depth exceeded threshold queue=webhooks depth=188

12:41:10apidatabase connection refused namespace=checkout-production

12:41:11victorialogsquery completed hits=7512 step=1m cache=warm

12:41:12hikari-apiquery helper prepared LogsQL for prompt="summarize checkout errors"

12:41:13ingressedge access authenticated service_token=automation status=ok

12:41:14kube-systemnode monitor heartbeat node=node-b.internal.example

12:41:15billinguncaught exception trace_id=01JZHKW8VX3PMD3KBA9

12:41:16schedulerjob delayed duration=31s reason=rate_limit

12:41:17hikari-uifacet selection namespace=edge-staging level=error

12:41:18hikari-mcpsummarize_window returned service hostname level namespace pod

12:41:19orders-apislow request method=POST path=/orders latency_ms=1432

12:41:20auth-servicetoken introspection failed issuer=legacy-auth status=503

12:41:21hikari-apifield aliases mapped observed service auth-service

12:41:22collectorbatch flushed bytes=82011 rows=500 duration_ms=66

Open source observability UI

Hikari

Shine a light on your logs.

Query VictoriaLogs with a focused interface, configurable facets, optional natural-language search, and MCP tools built for agents that need real operational context.

View on GitHub Explore MCP docs

Built for investigations

Readable logs, discoverable fields, bounded queries.

Search by intent

When OpenAI is configured, turn plain English into LogsQL with evidence, field discovery, and the exact query used.

Keep the UI close to the data

Facet by service, hostname, level, namespace, and pod before narrowing into raw events.

Agent-ready from day one

Expose the same investigation flow to MCP clients through tools for search, facets, hits, and summaries.

MCP endpoint

Give AI tools a controlled way to query logs.

Hikari ships MCP tools for window summaries, raw LogsQL, field discovery, facets, hit counts, bounded live-tail samples, and optional AI-assisted search.

https://logs.example.com/mcp

tools/list
  get_instructions
  summarize_window
  query_logs
  get_facets
  # ai_search appears when OpenAI is configured