Test design — not executedBrowser and computer agents

How to evaluate agents for web data extraction

Use a bounded task with observable results. The candidates below are research starting points, not tested recommendations.

Source-linked research and editorial guidance. Capabilities are not independently tested unless a profile explicitly includes a reproducible test record. Sources checked November 13, 2026.

Inputs
An allowed set of URLs, a required output schema and examples of missing or conflicting values.
Expected deliverable
Structured rows with source URLs, observation times and explicit missing values.
Procedure
  1. Define the schema
  2. Run on representative pages
  3. Retain source evidence
  4. Validate rows
  5. Inspect failures and retries
Approval boundary
Use permitted public pages or a fixture. Do not grant form submission or purchase rights for a read-only extraction task.
Acceptance criteria
Field accuracy, source coverage, duplicate rate, missing-value handling and cost per accepted row.
Failure modes
Guessing missing fields, mixing prices from different plans, losing provenance and treating a browser session as a completed extraction.

Candidates to investigate

Editorial topic relevance — task support not verified. Research starting points — none of these has been run against this task.

Sources

  1. browser-use.com (opens in a new tab)
  2. browser-use.com/pricing (opens in a new tab)
  3. ycombinator.com/companies/browser-use (opens in a new tab)
  4. browser-use.com/posts/seed-round (opens in a new tab)
  5. github.com/browser-use/browser-use/blob/main/CLOUD.md (opens in a new tab)
  6. github.com/browser-use/browser-use/blob/main/skills/open-sour… (opens in a new tab)
  7. skyvern.com (opens in a new tab)
  8. skyvern.com/pricing (opens in a new tab)
  9. docs.apify.com/mcp (opens in a new tab)
  10. browserbase.com (opens in a new tab)
  11. browserbase.com/pricing (opens in a new tab)

Something out of date? Suggest a correction