OBSERVABILITY PLATFORM

One platform for every log, error, and insight.

Sherlock is an observability platform built on ClickHouse. Explore logs, track errors, query with SQL or natural language, and let AI surface what matters.

level = 'ERROR' AND service LIKE 'api-%'Search

284.1K total127 errors2.3K warnings

datelevelmessage

14:23:01.443INFOGET /api/v2/users/me completed in 23ms

14:23:01.891INFOPOST /api/v2/orders/checkout session created

14:23:02.102WARNConnection pool reaching capacity (87%)

14:23:02.330ERRORNullPointerException in PaymentService.process

14:23:02.571INFORetry attempt 1/3 for upstream payment-gw

14:23:03.012INFOGET /api/v2/products?category=electronics 200

14:23:03.244INFOCache hit for session:a]f8c2e rate=94.2%

14:23:03.502WARNSlow query detected: 1,203ms on orders_idx

14:23:03.891INFOHealth check passed, all services green

14:23:04.102INFOWebSocket connection established client:8f2a

HOW IT WORKS

From your services to your screen.

Your applications emit telemetry via OpenTelemetry. Sherlock indexes it in ClickHouse and makes it queryable through a dashboard and an MCP server for AI assistants.

Interfaces

>_Dashboardlog explorer, error tracking

MCPAI AssistantsClaude, editors, CLI

Sherlock Platform

{ }Your Servicesapps, APIs, workers

OTel

OTelCollectorreceives, processes, exports

OTLP

CHClickHouseindex, store, query

APISherlock APIquery, search, alerts

LOG EXPLORER

Every log line, one search away.

Full-text search across your entire log volume. Filter by severity, service, time range, or any custom attribute. Drill into individual entries with a click.

>_Log Explorer

14:23:01INFORequest completed in 42ms

14:23:02WARNConnection pool near capacity (89%)

14:23:03ERRORNullPointerException in UserService.auth

14:23:03INFORetry attempt 1/3 for upstream call

14:23:04INFOHealth check passed, all services green

ERROR TRACKING

Errors grouped, deduped, and delivered to Slack.

Stop drowning in duplicate alerts. Sherlock automatically groups errors by pattern, deduplicates across time windows, and sends hourly digests to your Slack channels.

!!Error Groups

NullPointerException2,847

UserService.auth · first seen 2h ago

ConnectionTimeout1,203

Redis.getSession · first seen 6h ago

RateLimitExceeded412

PaymentGateway.charge · first seen 1d ago

QUERY ENGINE

From plain text to SQL. You choose the depth.

Start with a simple keyword search. Need more precision? Use SQL operators, dot notation for nested JSON fields, or combine filters with AND/OR. The query language grows with you.

SQLQuery Builder

Simple searchconnection timeout

SQL operatorslevel = 'ERROR' AND service LIKE 'api-%'

Dot notationhttp.status_code = 500

IN clauselevel IN ('ERROR', 'FATAL')

AGENT-NATIVE

Works for you. Works for your agents.

Sherlock exposes a full MCP server — so any AI agent can search logs, trace errors, and explain queries as part of its workflow. Give your agents the same observability tools your team uses.

MCPAI Tools

search_logsFull-text and structured search

get_error_groupsGrouped, deduplicated error patterns

explain_queryNatural language to ClickHouse SQL

count_logsAggregate counts by time bucket

get_log_detailsFull context for any log entry

list_sourcesAvailable data sources and schemas

GET STARTED

Ready to see what your logs are telling you?

Try Sherlock with your own data, or reach out to learn how it can fit your infrastructure.