An OpenTelemetry database Claude Code loves to query.
Point your coding agent at twelve months of unsampled spans and logs, already clustered and baselined for faster reasoning and fewer tokens. Connect to anything else using our Postgres URL.
› what's failing right now?
› why is checkout slower than last week?
› has this happened before?
› what changed after the deploy?
› create a dashboard showing what's new this week
Packrat stores your OTel data in a column store — a group-by over twelve months is instant — and clusters traces by workload and performance, which no GROUP BY can express. Together they let Claude reason deeper and faster during an incident.
Every OTel service gets its own read-only Postgres connection to its data, so anything that speaks Postgres reads it directly — Grafana, Metabase, pandas, ActiveRecord.
$ psql postgres://<role>@sql.packrat.dev/checkout
Team and above.
Typically 5–20× cheaper than Datadog, Honeycomb or Sentry, with 4–12× the retention.
free≤ 0.5 GB/day$0MCP tools, 12 mo, one project
team≤ 3 GB/day$49/mo+ Postgres URL, roles, ∞ projects
team≤ 10 GB/day$149/mo
team≤ 30 GB/day$399/mo
scale> 30 GB/daytalk to us+ SSO, own bucket, longer retention