Overview
Skrut runs grounded, multi-turn conversations against your live agent and grades every turn — so you get a readiness score you can defend, not a demo you hope holds up.Steps
1
Add your agent
Connect a voice agent via LiveKit, or a chat agent via your HTTP endpoint.
Skrut needs a URL it can send messages to and receive replies from.
2
Describe the expected outcome
Tell Skrut what your agent is supposed to do — in plain language.
It generates grounded test scenarios from your agent’s own knowledge base.
3
Run and get your readiness report
Hit Run. Skrut executes every scenario, grades each turn, and produces
a readiness report you can share with your team or re-run every release.