Niches · Observability & incidents
Search logs for a reported error or event
Two or three companies' docs show a full fit.
What counts as done
Matching log lines with timestamps and the query used, a count of total matches and a marker when results were truncated, from at most three queries.
How agents ask for it: search the logs for this request id
, find errors for this user in the last hour
, is this event in the logs
, what did the server log when it failed
Does the job 2
The docs show it meets the test above, and nothing we read says otherwise.
| Listing | Why |
|---|---|
| Datadog MCP Server D Datadog, Inc. | search_datadog_logs takes a query and time range, analyze_datadog_logs counts matches in SQL, and responses are truncated with instructions for fetching more. source |
| Grafana MCP Server BB Raintank Inc., d/b/a Grafana Labs | query_loki_logs takes LogQL, caps lines at --max-loki-log-limit (100) and requests one extra line to detect truncation, and Loki metric queries count matches. PII redaction is still an open request (#1124). source |
Does the job, with a limit 6
It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.
| Listing | Why |
|---|---|
| Axiom MCP Server BB Axiom, Inc. | APL search and count in one queryDataset call, but the docs state no result size cap and no truncation marker, and tool schemas are only visible after sign-in. source |
| Better Stack C Better Stack, Inc. | One SQL query returns matching lines and counts, and Plan query windows estimates rows read, but no result size limit or truncation marker is documented. source |
| Honeycomb MCP B Hound Technology, Inc. d/b/a Honeycomb | run_query is an aggregation query that returns counts and breakdowns; the docs show no raw-row result with a truncation marker, so matching lines come through traces and span details instead. source |
| New Relic MCP Server B New Relic, Inc. | list_recent_logs, get_log_statistics and NRQL over Log give lines and counts, but no result cap or truncation marker is documented and tool schemas need a signed-in session. source |
| Rootly MCP Server C Rootly Inc. | The observability toolset covers Rootly Observability logs and patterns, but the docs give no query syntax, match count or truncation marker for it. source |
| Sentry MCP B Functional Software, Inc. d/b/a Sentry | search_logs returns lines, counts and time series and accepts Sentry search syntax, but an embedded model translates every query, and no truncation marker is documented. source |
Doesn't do this job 4
Listings in the category that do something else, with the reason.
| Listing | Why |
|---|---|
| Causely MCP server E Causely, Inc. | get_logs returns live or diagnosis-time logs for one entity; there is no search syntax, match count or truncation marker. source |
| Cronitor MCP server C Cronitor | The docs call the server a control plane, not a data plane; there is no log search. source |
| incident.io API B Pineapple Technology Ltd. | No log query tool; ask_telemetry returns a synthesised answer with no documented match count or truncation marker. source |
| PagerDuty MCP Server E PagerDuty, Inc. | No log search; browse_activity returns PagerDuty's own log entries only. source |
What stops agents finishing
- empty result returned where the query failed
- no cap on range or bytes, so one query floods the context
- personal data in logs that policy bars from the model provider
Around it
Before: support.ticket-received, observability.investigate-alert. After: observability.error-to-fix, code.open-pr.
Where this comes from
Weight 16 of 100 in Observability & incidents: 8 of 42 accounts (Better Stack, Axiom, CloudWatch, Loki through Grafana)
The same as data: /api/v1/use-cases/observability.search-logs.json, or the MCP tool get_use_case with id observability.search-logs. Other jobs in this category: /api/v1/niches?category=observability.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.