Skip to content

Concepts overview

Read these pages in order to understand the runner-first model.

  1. Mental model — workflow and ownership
  2. Native model and framework runners
  3. Framework runners
  4. Data model
  5. Evidence and trajectories
  6. Scores and gates
  7. Annotation — labeling captured sessions (span / trace / session)
  8. Environment bundles and dependency tapes — executable worlds for agent eval
  9. How it works

By persona

Framework owner

TopicPage
First runQuick start
Runner packagingPrepare a framework run
Promptfoo specificsPromptfoo integration
Runner semanticsFramework runners

Platform operator

TopicPage
Trust boundariesTrust boundaries
Storage and retentionStorage and thelake
Runtime modelKernel and hosts

CI / release owner

TopicPage
Compare and promoteCompare and promote
Result statusResult status
EventsEvents

Key principle

Softprobe owns the outer workflow and environment, frameworks own inner evaluation semantics.

text
framework suite + subject + environment + runner
→ WorkflowVersion → framework runner
→ native result bundle + evidence
→ Softprobe compare/gate/governance

Zero code changes · Full-context visibility · Cost optimization