What evidence should a regression investigation include?

A finished investigation names the specific change responsible, not just the fact that something regressed, and it carries three things to back that up. First, the cause itself, stated as a located change: a commit, a prompt edit, a model swap, or an upstream system, not a vague description of symptoms. Second, the failing traces and findings that actually support naming that change, so the claim is checked against evidence rather than a hunch. Third, the hypotheses you tested and ruled out along the way, because a cause that survived elimination is a stronger claim than the first plausible guess. Skip the ruled-out list and the write-up reads like a guess even when the conclusion is right, since nobody reading it can tell whether other candidates were considered. The evidence also tells you where the fix belongs: a cause that’s a located change points straight at the place the fix has to land.

keep reading

More on this.

Two ways to run Tessary.

Tessary is an open-source agent reliability platform. Cloud and self-hosted run the same workflow on the OpenTelemetry traces your agent already emits.

Tessary Cloud

We host it for you. Send your first trace with nothing to deploy and no model key.

what's includedper organization
traces
10,000 per calendar month
stored trace data
1 GB
retention
30 days
model credit
$10, one-time, for triage and root-cause analysis
credit card
not required

Self-hosted Tessary

Run the open-source code on your own infrastructure with one command. Add your own model key for triage and root-cause analysis.

Self-host Tessary for me by following https://github.com/tessaryai/tessary/blob/main/setup.md

docker compose -f oci://docker.io/tessaryai/tessary:compose up -d -y