Skip to main content

Branch QA Report

The Branch QA Report validates a branch before it merges. GetWebstack generates test cases from the diff, runs them against your full app in isolated sandboxes, and returns a shareable report: pass/fail per case, repro steps, and the suspected failing code area.

It builds the environment straight from your code, so it can control the seed data and run many isolated sandboxes at once. A QA tool that only gets a URL can't.

Prerequisites​

1. Generate test cases​

In your AI agent chat, on the branch you want to validate:

/gws-qa-generate

The /gws-qa-generate skill reads what the branch changed (vs. the default branch), maps it to user-reachable flows, grounds each case in the branch's intent (spec docs, PR/ticket acceptance criteria), and seeds the cases to GetWebstack.

If the branch has a spec .md but no implementation yet, generation runs spec-first: the cases become the contract the implementation must pass.

2. Review and edit​

Add, remove, or sharpen the test cases in GetWebstack dashboard. Decide what "correct" means before anything runs.

3. Run them in parallel​

Back in your AI agent chat:

/gws-qa-run

The /gws-qa-run skill runs each case in its own isolated gws fork (read-only cases share one), against the full stack with controlled seed data. Once finished, it publishes the results to the dashboard.

4. Read the report​

Open the QA Run Report in the dashboard: pass/fail per case, evidence, repro steps, the suspected failing code area for failures, and how the results map to the branch's acceptance criteria.

Share the link with reviewers and decide whether the branch is safe to merge.

See also​