Skip to content

Runs

A run is the unit of test execution that one or more sessions belong to. Runs exist so you can group, search, and bill all the work an agent did for a single goal — a CI build, an exploratory test, a one-off probe.

  • Auto-create (default) — kifas.session.create without a run_id lazily creates a run and assigns the session to it. The run inherits the project from the API key.
  • CI-supplied (Slice J) — your CI orchestrator passes a stable run_id so all sessions in a single CI execution land in the same run. (Available when Slice J’s CI integration ships.)
Run
├── status: pending | running | passed | failed
├── created_at, ended_at
├── source: { ci?: { provider, commit, branch }, ... }
├── sessions: [Session, Session, ...]
└── artifacts: [Recording, Screenshot, Trace, ...]

Runs are scoped to the API key’s organization. The dashboard surfaces every run for every member of that org with read access; RLS policies enforce the boundary.