Suites & steps

A suite is the durable artifact WeftKit produces from a recording: an ordered list of semantic steps, each with the locators, waits, and assertions needed to reproduce it. You edit intent here — never implementation code.

The WeftKit suite editor showing a versioned suite with five replay steps, each labelled with its action type and assertion count, and a step inspector on the right. — The suite editor. Each step shows its action type, executable assertion count, and locator count.

What a step contains

Fields of a WeftKit suite step
Field	Meaning
Step name & intent	The human description of what this step accomplishes, kept separate from how it is performed.
Action	navigate, click, fill, select, press, tab, or assert.
Locator fallback ladder	Ordered strategies for finding the target element. Replay walks the ladder before it gives up.
Timeout	An explicit wait budget, in milliseconds.
Tab alias	Names the tab a step acts on, so multi-tab flows replay against the right window.
Risk class	R0 to R4 — how consequential this action is if it executes for real.

Risk classes

Every step carries a risk class, and that class drives the policy applied at replay. Read-only navigation and inspection sit at the bottom of the scale; payments, irreversible deletion, publication, and outbound messages sit at the top and require explicit handling rather than silent execution.

Risk classification is a product invariant, not a preference. Consequential actions require policy no matter which capture profile or provider you have configured.

Executable assertions

Steps are counted by their executable assertions — the checks replay will actually evaluate — so the editor never inflates coverage with decorative expectations. An assertion recorded into a revision cannot be weakened afterwards: an AI repair proposal has to validate against the immutable revision, not replace it.

Revisions are immutable

A suite revision is frozen once saved. Editing produces a new revision rather than mutating the one your past runs were executed against, which is what makes an old report still mean something. The editor shows the current revision, and the revision history tab records how the suite got here.

The revision history tab of the WeftKit suite editor, listing suite revisions in order. — Revision history. Past runs stay interpretable because the revision they ran against never changes.

Suite status

Ready — the suite has executable assertions and every prerequisite is configured, so it can be replayed.
Needs attention — something blocks a faithful replay, most commonly a secret reference that has no value configured yet in the vault.

LimitationA suite that reads a one-time code cannot be fully automated. Those references stay human checkpoints by design and are never accepted into the vault.

Next: Secret vault