Why is my agent taking an action it never took before?

Either your agent’s logic changed on purpose, a new branch, a new tool, a prompt update opening a path it didn’t have, or something is pushing it somewhere it was never designed to go. Either way, it’s now doing something outside its own history at that call site.

A brand new action is the easiest kind of drift for Tessary’s behavior_drift classifier to catch: a tool or step your agent has never used before stands out clearly against a baseline built from what it has actually done. What the classifier can’t tell you on its own is whether that’s fine. A deliberate change needs to recur across enough sessions before Tessary treats it as the new normal instead of flagging it every time; a real bug needs a person to look at the case triage opens and rule it isn’t.

Until one of those happens, expect it to keep showing up as a finding.

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