>_
grepd_
$ install grepd
erik@workstation — grepd tail — 132×38
$ grepd tail prod/*.log --since 15m --grep 'timeout|retry'
→ 4 sources · 1,204,881 lines scanned · 38 matches · 214ms · streaming

Every log line.
One grep.
No index.

Point grepd at anything that writes lines — files, journald, Kubernetes, an S3 bucket — and search it live from your own terminal. No agents to deploy, no index to rebuild, no per-gigabyte bill at the end of the month.

curl grepd.sh | sh [ read the docs ] 12 MB static binary · macOS, Linux, BSD
ctrl-r fuzzy history · /pattern to refine in place · :w to pin a view · q to quit
$
grepd 0.9.4 0:tail* 1:query 2:alerts us-east · 14 nodes12:04:38
# benchmark · 1 TB of nginx + application logs, cold, single laptop — bars are grepd against the hosted average
0.9s
median query, 1 TB cold
Hosted stacks average 41s for the same scan.
38MB
resident memory per host
Agent-based collectors sit at roughly 410 MB.
0 daemons
one binary, no sidecar
Drop it in $PATH. It reads what you already write.
$39/mo
flat, per team, unmetered
Ingest as much as you like. The bill never moves.