PLAYWRIGHTMCPAGENTSMERGE QUEUEFLAKE QUARANTINECI GATES
ben@spurlock:~$

BEN
SPURLOCK

Senior test automation consultant in Salt Lake City. I help Series A–C startups (10–50 engineers) build Playwright systems that scale without breaking.

ben@spurlock:~$
whoami
ben spurlock — sdet since 2015 — web since 2012
ben@spurlock:~$
10+
years SDET
13+
years web
5
teams unified
ben@spurlock:~/services$

ls ./offers

init-playwright.sh

Playwright from scratch

I design frameworks your team can own: fixtures, page objects, and reporting that survive the second team using them.

wire-ci.sh

CI that does not lie

Merge queues, sharding, and flake quarantine so a green build means something again.

ai-loop.sh

Agents in the test loop

MCP access to the suite, agents that patch broken tests, and reports that separate real failures from flake.

more in /services →

ben@spurlock:~/ai$

ls ./ai

AI is useful in a test suite when it can run the same commands you run. I put MCP and agents on the framework, not in a slide about “AI-powered QA.”

mcp.sh

MCP on the test repo

Wire Cursor, Claude, and other clients to Playwright, traces, and CI so an agent can run a spec, read the failure, and work from the same evidence you do.

faster.sh

Faster development

Agents draft specs from tickets and diffs. You review, tighten selectors, and land the coverage instead of starting from a blank file.

repair.sh

Repair broken tests

When a locator or wait rots, an agent opens a PR with a fix and the trace that proves it. Humans still merge.

triage.sh

Report real failures

Triage bots that stop paging you for flake and file the product bugs — with repro, screenshot, and the last good commit.

mail

ben@spurlock:~$ cat next-steps.txt

Let's discuss your testing challenges

Tell me about flake, pipeline time, or the framework nobody wants to own. I will tell you if I am the right person.

./talk.sh

ben@spurlock:~$

Senior test automation consultant in Salt Lake City. I help Series A–C startups build Playwright systems that scale without breaking — including MCP and agents in the test loop.

© 2026 Ben Spurlock · phosphor tty · logout