browser-task-artifact

name	browser-task-artifact
description	Turn a repeatable browser or web task into a verified reusable artifact with task notes, critical points, a final script, logs, screenshots, structured results, and verification evidence.

Browser Task Artifact

Purpose

Use this skill when a browser task needs durable proof: a checked workflow, a reusable script, evidence screenshots, action logs, or structured results. This is different from baseline route smoke. The browser is disposable; the artifact workspace is the state.

Use playwright-smoke-setup instead when the user only wants Chromium route smoke tests, playwright.config.ts, GitHub Actions wiring, or route marker maintenance.

Output Contract

Create one workspace under .tmp/browser-task-artifacts/<task-id>/ unless the user or repository policy names another location.

Prefer the package CLI:

npx playwright-ui-smoke-kit artifact-init \
  --task-id <task-id> \
  --title "<human-readable title>" \
  --source "<url-or-local-route>" \
  --script-ext ts

Before treating the workspace as proof, run:

npx playwright-ui-smoke-kit artifact-check .tmp/browser-task-artifacts/<task-id>
npx playwright-ui-smoke-kit artifact-check .tmp/browser-task-artifacts/<task-id> --strict

The workspace should contain:

• task.md with the original task, sources, assumptions, and secret-handling notes.
• plan.md with critical points that must be proven.
• final_runs/run_<n>/final_script.* with the clean reusable script.
• final_runs/run_<n>/action_log.md or action_log.jsonl.
• final_runs/run_<n>/result.json with structured output.
• final_runs/run_<n>/screenshots/ with evidence screenshots.
• verification.md mapping each critical point to concrete evidence.

Workflow

1. Read the target UI or site from current evidence; do not rely on memory.
2. Fill task.md and plan.md before coding.
3. Explore with scratch code inside the artifact workspace.
4. Create or replace final_runs/run_<n>/final_script.*.
5. Run the final script from a clean start, preferably through:

npx playwright-ui-smoke-kit artifact-run .tmp/browser-task-artifacts/<task-id> -- [args...]

6. Save action logs, screenshots, and result.json in the selected run directory.
7. Update verification.md so every critical point cites concrete evidence.
8. Run artifact-check --strict before reporting the artifact as verified.

Reuse Rules

When the task has variable inputs, parameterize the final script with CLI flags. The script should have safe defaults that reproduce the original task, avoid side effects at import time, and show --help when reuse matters.

Safety

• Ask before running workflows that mutate external state, submit forms, make payments, send messages, or post publicly.
• Never store secrets, bearer values, private keys, auth headers, full .env contents, or private customer data in committed artifacts.
• Commit only sanitized summaries or deterministic example artifacts.
• Promote an artifact into CI only after it is deterministic, safe, and fast.