Skip to main content

Trace capture capabilities and limits

Help & Support

Answers for engineering and platform teams evaluating Versalist for agent evaluation and skill revision.

How do teams typically start?

Start with a pilot on one workflow: define the test cases, compare the current version of your agent’s skill with the new one, and review the results with our team. Scope, availability and fees are agreed before work begins.

How are agent runs evaluated?

A challenge scores each run against its test cases and rubric, per case and per dimension. When trace capture is enabled, the run also records bounded call metadata. Coverage depends on the execution path and configuration. See /docs/trace-capture for supported events and limits.

What does private challenge access mean?

Discuss private challenge libraries and controlled team access during scoping. This does not establish a private VPC, on-premises deployment or customer-hosted platform. Deployment options at /deployment shows the status of each. Check Trust for unresolved procurement questions.

Still have a question?

Request a demo for rollout and pricing questions, or email support for account and product help.