bronto is installed and you have a
management API key from Settings > API Keys in the Bronto web app. The examples use a
small demo org: a checkout-service in the prod collection that started returning
502 errors because calls to payments-api time out.
The five-minute path
1. Log in
--key-stdin form.
2. Check the connection
OK — https://api.eu.bronto.io (42ms). If it fails, run bronto auth status to see which
key and region the CLI picked up.
3. See what data you have
-d/--dataset, you can pass a dataset name instead of its id.
A name that exists in more than one collection needs the collection in front of it, like
prod/checkout-service.
If you mostly work with one dataset, make it the default and drop -d:
4. Run your first search
message_kvs., so status in the query is the message_kvs.status field in the output.
To see every field of a single event, untruncated, add -x:
bronto fields lists the fields in a dataset with sample
values:
5. Pipe it
When stdout isn’t a terminal,search prints JSON Lines (one JSON object per event), so
you can pipe it into jq, wc, or a script. Each event is flattened into dotted keys:
--fields keeps only the keys you name. A bare name such as status finds
message_kvs.status, and the output key is the name you typed. --jq runs a jq expression
on each result without needing jq installed, and -r prints strings without quotes:
--fields, --jq sees the whole event, so it uses the full message_kvs. key
names. With --fields, --jq runs on the narrowed result and sees the names you typed:
--fields host --jq .host.
Where next
- Searching: filter flags, aggregates, histograms, patterns, and the interactive REPL.
- Live tail: follow new events as they arrive.
- Traces: find the slow
payments-apispan behind those 502s. - Scripting and agents: output formats, exit codes, and jq recipes.

