// Shopping
Compare laptop specs across product pages
Search Staples, open the top three laptops and normalize their specs into one table.
11 agent tool calls, click to jump
Why it’s hard
Each product page hides the numbers you need behind a tab, in a slightly different layout.
- Specs sit behind a tab below the fold that only renders once opened
- Sponsored tiles are mixed into the search results
- Only the first row of results renders until the page is scrolled
Stagehand vs. Playwright MCP
Given the same goal and model, the Stagehand code-mode agent finished for 74% less than an agent driving Playwright MCP.
- stagehand code mode agent2/3 runs succeeded
- $0.163
- playwright mcp agent3/3 runs succeeded
- $0.617
medians over successful runs, or over all runs when none succeeded; lower is better
Then run it as a script
Once the flow works, the cookbook runs it as plain Stagehand calls. With caching on, a re-run cost about the same as the first run.
- cookbook, first run3/3 runs succeeded
- $0.211
- cookbook, cached re-run3/3 runs succeeded
- $0.215
medians over successful runs, or over all runs when none succeeded; lower is better
anthropic/claude-sonnet-5 at $2 input · $0.2 cached · $2.5 cache write · $10 output per 1m tokens · stagehand 4.1.0 · playwright mcp 0.0.82 · agents stop after 60 llm calls · run on 2026-09-25 · methodology
Run it yourself
Give the same goal to Claude Code with Stagehand’s code-mode MCP server, which exposes three tools: run, snapshot and screenshot. Once the flow works, the cookbook runs it as a plain Stagehand script.
Search Staples for "laptops 16gb ram". Open the first three laptops in the results that are not sponsored and, for each, report the name, current price in USD, processor, RAM in GB, storage in GB, screen size in inches, weight in pounds (null if not listed) and the product page URL.