NTM Browser Test Coordination

Use this skill when browser or UI validation needs more coordination than a single local test command: multiple scenarios, parallel panes, visible browser evidence, failure triage, or a handoff that another operator can resume.

This skill does not replace the browser test runner. It defines how the lead agent scopes scenarios, assigns NTM panes, requires evidence, records failures, and returns one verifiable verdict.

Coordination Rules

• Scenario first. Define the user-visible scenario before dispatching panes. A command without a scenario is only a test invocation, not a UI validation plan.
• One pane owns one scenario slice. Each pane receives a scenario id, fixture slice, command, evidence path, and completion contract. Avoid shared users, shared browser state, and overlapping writes.
• Evidence beats narration. A pane saying "passed" is not enough. Require screenshots, traces, videos, console logs, test reports, or a specific "no artifact because" explanation.
• Failures become records. Every failed or blocked scenario gets a failure record with reproduction command, artifact paths, expected behavior, observed behavior, and current owner.
• Handoff is part of the run. The run is not complete until the lead can state what passed, what failed, what was not run, where evidence lives, and what the next operator should do.

Workflow

1. Define the run

Create a coordination packet before touching NTM: