Skip to main content

/e2e-from-recording (primary)

What it does​

/e2e-from-recording is the primary path for turning a captured browser flow into a production-quality test. It chains three agents in sequence:

  1. Stabilizer — cleans the raw VS Code recording and paired English steps into a canonical spec file (tests/specs/) plus data entries (tests/data/).
  2. Generator — converts that spec into an executable .spec.ts file under tests/e2e/, then smoke-runs it.
  3. Reviewer — checks the generated test against the project rulebook (tests/agents/rules/playwright-rules.md) and emits a PASS or FAIL verdict. If the test fails review, the pipeline loops generate → review at most once before handing control back to you.

When to use it​

Use /e2e-from-recording whenever you have already captured a flow in the VS Code Playwright Recorder and saved it as a .spec.ts file in tests/recordings/. This is the fastest, most reliable path from a captured flow to a committed test.

If you are designing a test from scratch without a recording, use /e2e-from-scratch instead.

How to run it​

/e2e-from-recording tests/recordings/add-billing-contact.spec.ts

On Antigravity, the slash command takes no inline argument — run /e2e-from-recording, then provide the recording path when prompted.

Example​

Input:

  • tests/recordings/add-billing-contact.spec.ts — the raw Playwright Recorder output.
  • tests/recordings/add-billing-contact.steps.md — the paired English steps file describing the intent of each action in plain language.

What you get back:

  1. A canonical spec at tests/specs/billing/add-billing-contact.md (Stabilizer output).
  2. A runnable test at tests/e2e/billing/add-billing-contact.spec.ts (Generator output).
  3. A smoke-run result showing whether the test executes cleanly in the target environment.
  4. A Reviewer verdict — PASS if the test meets all project standards, FAIL with line-precise findings if something needs fixing.
  5. An alignment report at tests/recordings/.stabilizer-reports/add-billing-contact-alignment.md mapping each raw recorder line to a logical step.

Notes​

  • The pipeline does not commit. You review the output and commit when satisfied.
  • The paired English file is required. The Stabilizer halts with an error if tests/recordings/<name>.steps.md is missing or if the steps are too vague to map to the recording. Write at least one plain-English sentence per logical action before running the command.
  • Smoke-run limit. If the generated test fails to run three times in a row, the pipeline halts and reports the failure rather than looping indefinitely. Fix the underlying issue (usually a selector or data problem) before retrying.
  • Generate → review loop. If the Reviewer returns FAIL, the Generator revises and the Reviewer runs once more. After two review cycles the pipeline stops and presents the findings to you.