Playwright MCP
Tied for most recommended of 21 MCP servers
Lets AI agents control a browser to test front ends and check their own work. Critics say it is slow and uses many tokens.
Lets AI agents control a browser to test front ends and check their own work. Critics say it is slow and uses many tokens.
62 people gave an opinion
- recommend it
- 40
- criticize it
- 22
- approval share
- 65%
Recommended in 11 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 40 people recommend itNeeds 10 or more, met.
- 1.8 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 11 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Tests front end work well
“I've had pretty good luck with using playwright mcp to test web front ends.”
Comment from 14 May 2026 in “God Damn AI is making me dumb” “The only useful MCP for me is Playwright for front end work.”
Comment from 16 Nov 2025 in “What if you don't need MCP at all?” “which lets something like Claude directly test a website”
Comment from 5 May 2026 in “Computer Use Is 45x More Expensive Than Structured APIs” Lets agents check their own work
“You need to give it the tools to check its own work”
Comment from 16 Feb 2026 in “Ask HN: Why is my Claude experience so bad? What am I doing wrong?” “the second is giving the model a feedback loop”
Comment from 2 Feb 2026 in “Two kinds of AI users are emerging. The gap between them is astonishing” “At most get playwright MCP in so the agent can see the rendered output”
Comment from 28 Mar 2026 in “Anatomy of the .claude/ Folder” Works well in daily use
“I know the Playwright MCP server works great. I use it daily.”
Comment from 14 Dec 2025 in “Ask HN: How can I get better at using AI for programming?” “playwright MCP often works very well”
Comment from 6 May 2026 in “Computer Use Is 45x More Expensive Than Structured APIs” “has worked pretty well for me”
Comment from 15 Mar 2026 in “Let your Coding Agent debug the browser session with Chrome DevTools MCP” Parses pages and gets tasks done
“i use playwright mcp to do a bunch of things and then make llm dump code so that i can do something in a reproducible manner”
Comment from 4 Nov 2025 in “MCP is the wrong abstraction [video]” “tools like playwright MCP can easily parse a webpage to use it and get things done with existing markup today”
Comment from 17 Feb 2026 in “WebMCP Proposal” “Playwright will just do stuff.”
Comment from 16 Sep 2026 in “Mistral X Mozilla: Private, Multilingual AI Browsing”
What people warn about
Each mark is one comment
Slow and uses many tokens
“Aside from being the slowest tool calls I have ever seen, the agent struggles mightily to figure out the simplest of navigation and interaction.”
Comment from 9 Feb 2026 in “OpenClaw Is Changing My Life” “A single Playwright MCP server burns 11.7K tokens on tool definitions alone, included in every request whether you use them or not.”
Comment from 9 Feb 2026 on Hacker News “Each call dumps the entire page accessibility tree into the context window.”
Comment from 21 Feb 2026 in “OpenBrowser-AI MCP” - Comment from 17 Feb 2026 in “Show HN: PageMap – MCP server that compresses web pages from 100K to 2-5K tokens”
- Comment from 18 Jun 2026 in “Launch HN: TesterArmy (YC P26) – Agents that test web and mobile apps”
- Comment from 22 Feb 2026 in “I made my favorite AI tool”
Unreliable and breaks often
“the playwright MCP is very unreliable and inefficient to use”
Comment from 6 Feb 2026 in “Show HN: Smooth CLI – Token-efficient browser for AI agents” “it assumes the agent can reason about CSS selectors and DOM state, which breaks constantly”
Comment from 7 Feb 2026 in “Show HN: Smooth CLI – Token-efficient browser for AI agents” “I always have this problem with Playwright MCP on browser”
Comment from 20 Aug 2026 in “Show HN: MCP app for Android, drive apps via AI (no root, PII redacted locally)” - Comment from 21 Dec 2025 in “Claude in Chrome”
Only one instance or tab
“I had an issue with playwright MCP where only one Claude Code instance could be using it at a time”
Comment from 27 Mar 2026 in “Anatomy of the .claude/ Folder” “it's a fundamental limitation of that MCP that only on instance/tab can be controlled”
Comment from 28 Mar 2026 in “Anatomy of the .claude/ Folder” Spawns headless browser instances
“Official MCPs (Playwright and Chrome dev tools) spawns new browser instance in headless mode”
Comment from 8 Nov 2025 on Hacker News “for some reason claude in headless mode could not open playwright MCP in headless mode”
Comment from 7 Nov 2025 in “You should write an agent”
Other MCP servers people recommend
The four most recommended other MCP servers, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| Playwright MCPThis page | 40 | 22 | 65% | 11 | — | — |
| Context7 | 40 | 7 | 85% | 10 | — | Approved |
| Chrome DevTools MCP | 36 | 22 | 62% | 9 | — | — |
| Serena MCP | 18 | 3 | 86% | 9 | — | Approved |
| Context Mode | 16 | 1 | 94% | 6 | — | Approved |
What people said
40 of the 62 comments that mention it, newest first. Each links to the original comment.
“Playwright will just do stuff.”
“I always have this problem with Playwright MCP on browser”
“it easily consumes 1M+ tokens for a test with ~20 steps”
“I've had pretty good luck with using playwright mcp to test web front ends.”
“playwright MCP often works very well”
“which lets something like Claude directly test a website”
“it's a fundamental limitation of that MCP that only on instance/tab can be controlled”
“At most get playwright MCP in so the agent can see the rendered output”
“I had an issue with playwright MCP where only one Claude Code instance could be using it at a time”
“you also should take a look at https://github.com/microsoft/playwright-mcp”
“I have some a Gitea/Playwright mcp/webservice setup, which works well with both.”
“Every single one defaults to a safe extraction method”
“has worked pretty well for me”
“Very little has value add over just being able to control a browser”
“I was happy with playwright like MCPs that require the daemon so didn't convert them to CLIs”
“It's not the fastest thing in the world because it still has to spin up a browser”
“Each call dumps the entire page accessibility tree into the context window.”
“dumps the entire accessibility tree per page, which is 50-540K tokens”
“tools like playwright MCP can easily parse a webpage to use it and get things done with existing markup today”
“You need to give it the tools to check its own work”
“I am sure you must already know about Playwright mcp so why this?”
“A single Playwright MCP server burns 11.7K tokens on tool definitions alone, included in every request whether you use them or not.”
“Aside from being the slowest tool calls I have ever seen, the agent struggles mightily to figure out the simplest of navigation and interaction.”
“it assumes the agent can reason about CSS selectors and DOM state, which breaks constantly”
“the playwright MCP is very unreliable and inefficient to use”
“the second is giving the model a feedback loop”
“Adding playwright MCP changed my CC from mediocre frontend skills, to really really good.”
“it burned through like 150 premium requests in 5 minutes”
“you could have made if much simpler using playwright mcp.”
“we've already crossed the Rubicon with tools like Claude Code and Playwright MCP.”
“Yolo Mode with MCP Playwright and MCP Google Chrome Debug is already sudo on my system”
“After a lot of trouble trying to get playwright mcp to work on Linux”
“they recommended to try out something like Playwright MCP”
“I know the Playwright MCP server works great. I use it daily.”
“the only Microsoft AI product that my colleagues actually cared and use”
“The only useful MCP for me is Playwright for front end work.”
“the Playwright MCP server provided what I needed for now”
“Official MCPs (Playwright and Chrome dev tools) spawns new browser instance in headless mode”
“for some reason claude in headless mode could not open playwright MCP in headless mode”
“i use playwright mcp to do a bunch of things and then make llm dump code so that i can do something in a reproducible manner”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.