Axiom for Splunk app commands

Generating commands must be the first command in a Splunk search, so they start with a leading pipe. axlookup is a streaming command and appears later in an SPL pipeline, after a base Splunk search. axtimechart generates reporting output, so its results are a chartable table rather than events.

Time ranges#

Commands that query Axiom data use the Splunk dispatch time range by default. The Splunk time picker, saved search schedule, or dispatch earliest and latest values become the Axiom query window. A user who runs a search for Last 15 minutes gets the same 15-minute Axiom window without adding command-specific time arguments.

This applies to axsample, axsearch, axstats, axtimechart, axlookup, and axquery. axdatasets and axfields don’t query event time ranges.

Each boundary resolves independently:

If a boundary can’t be resolved, that side is omitted from the Axiom request. Splunk All time is treated as unbounded.

Use explicit options when the Axiom query window should differ from the current Splunk search range:

| axsearch dataset="http-logs" q="error" startTime="-24h" endTime="now" limit=100

Supported time inputs:

• RFC3339 UTC, such as 2026-07-02T12:00:00Z
• Splunk-style relative time, such as -15m, -24h, -7d@d, @d, or +1h
• now

Relative inputs resolve on the Splunk search head at command runtime and are normalized to UTC before being sent to Axiom. If an exact boundary matters, use explicit RFC3339 UTC.

To inspect the resolved range without querying Axiom, use explain=true on axsearch, axstats, or axtimechart:

| axsearch dataset="http-logs" q="status>=500" explain=true
| table compiled_apl, startTime, endTime

Splunk-like q syntax#

axsearch, axstats, and axtimechart support a constrained Splunk-like q option that compiles to APL. Supported patterns include:

q="error timeout"
q="error OR timeout"
q="error NOT debug"
q="(error OR timeout) service=api"
q="status>=500"
q="service=checkout"
q="service!=checkout"
q="trace_id=*"
q="service IN (api, worker)"
q="service NOT IN (api, worker)"
q="message:timeout*"

Use the where option on axsearch, or use axquery, when you need an APL predicate outside this subset. To see the APL a q expression compiles to, add explain=true.

Field conventions#

Every command returns Axiom fields as real Splunk fields. You don’t need to append | spath to reach them:

| axsearch dataset="http-logs" limit=10
| table _time service.name severity_text body

• Nested Axiom objects are flattened to dotted field names, so attributes.http.method is addressable directly.
• Arrays arrive as Splunk multivalue fields.
• Commands other than axtimechart keep the original row in _raw as JSON, so the Events tab shows the underlying Axiom record.
• axsample, axsearch, axstats, and axtimechart add axiom_dataset to each returned event. axlookup adds axiom_lookup_dataset to enriched events.

axlookup output fields are prefixed and normalized to Splunk-friendly names. The default prefix is axiom_, and punctuation in Axiom field names becomes _:

The app also ships Splunk workflow actions. When an event has axiom_dataset plus a supported field such as trace_id, span_id, service, or service.name, the event menu can launch a focused Axiom search for traces, spans, services, or Kubernetes deployments. The action preserves the current Splunk time range and opens the search in Axiom.

Command reference#

axdatasets#

List datasets visible to the configured Axiom token.

| axdatasets
| table id, name, kind, retentionDays
| sort name

axfields#

List fields for one Axiom dataset.

| axfields dataset="http-logs"
| table name, type, unit
| sort name

axsample#

Fetch recent events from one dataset without writing a query.

| axsample dataset="http-logs" fields="service,status,message" limit=20

axsearch#

Search one Axiom dataset and return event-shaped rows.

| axsearch dataset="http-logs" q="status>=500 service=checkout" fields="service,status,message" limit=100

axstats#

Push grouped aggregations into Axiom.

| axstats dataset="http-logs" q="status>=500" stats="count as errors, avg(duration_ms) as avg_duration" by="service" limit=100

Supported aggregation functions:

• count or count()
• sum(field)
• avg(field)
• min(field) and max(field)
• dcount(field) or dc(field)
• stdev(field)
• variance(field)
• percentile(field, pct), perc(field, pct), or p95(field) style shorthand

Use axquery for APL aggregations outside this subset, such as conditional aggregations.

axtimechart#

Push time-bucketed aggregations into Axiom and return a time series table.

This is a reporting command. The result is a stats table with _time plus the aggregation columns, so the Visualization tab works without further piping.

| axtimechart dataset="http-logs" q="status>=500" span=5m agg="count as errors" by="service" limit=1000

axlookup#

Look up the latest matching context from Axiom and add selected fields to incoming Splunk events. It follows Splunk’s lookup mental model, but the backing source is an Axiom dataset rather than a Splunk lookup table.

index=main service=api
| axlookup dataset="deployments" on="service=service.name" fields="version,owner,team"
| table _time, service, axiom_version, axiom_owner, axiom_team

The time range controls the Axiom lookup dataset, not the already-returned Splunk events. By default, the same Splunk dispatch window applies to both. Override the lookup window when your context data lives outside the base search window.

axquery#

Run raw APL when you need full control of the Axiom query language.

| axquery apl="['http-logs'] | where status >= 500 | summarize errors=count() by service, bin(_time, 1h) | sort by _time asc"
| table _time, service, errors

Pagination notes:

• Without pagination, Axiom’s per-page limit applies.
• With pagination, put | limit n in the APL to cap total returned events.
• paginate=cursor uses Axiom cursors. paginate=time advances by _time and deduplicates rows.

For the APL language itself, see the APL introduction and the Splunk SPL to APL cheat sheet.

Debug searches#

For axsearch, axstats, and axtimechart, start with explain=true. It shows the generated APL and resolved time range without querying Axiom, which separates app syntax issues from Axiom query issues:

| axstats dataset="http-logs" q="status>=500" stats="count as errors" by="service" explain=true
| table compiled_apl, startTime, endTime

Use debug=true with axquery when you need detailed runtime logs, then inspect the latest run with the axquery_latest search macro:

`axquery_latest`