frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Browser DevTools MCP: 78% less tokens than Playwright MCP, faster and consistent

https://medium.com/@serkan_ozal/browser-devtools-mcp-78-fewer-tokens-vs-playwright-mcp-faster-and-more-consistent-32f314004d30
3•sozal•1h ago

Comments

sozal•1h ago
We ran the same verification task with Claude Code, same prompt, 3 runs per MCP. Browser DevTools MCP: ~330K tokens, 12 turns, ~2m45s. Playwright MCP: ~1.5M tokens, 48–52 turns, ~6m. So 78% fewer tokens, 76% fewer turns, ~57% faster. We compare task-level totals rather than single-turn context size, and the post explains the methodology and how we designed for fewer round-trips (execute tool, scoping). Link in the post.
heisen_berg•1h ago
Great write-up, but looking at the results, it seems the massive 78% token reduction comes primarily from the execute tool batching actions, rather than DevTools being inherently lighter than Playwright. If someone added a similar sandboxed execute tool to Playwright-MCP to allow batching, wouldn't the token usage and turn count be nearly identical? I'm trying to understand if the protocol (CDP) is the real winner here, or just the smart API design of your MCP
sozal•1h ago
Yeah, you’re right. The most of the gain is from the execute tool (batching many steps in one call), not from CDP vs Playwright. If Playwright MCP added a similar sandboxed “run script that calls other tools” and the model used it the same way, token use and turn count would likely get much closer. So the win is API design (batching + scoping + lean tool defs), not the protocol. The post is about showing that this kind of design matters, we’re not claiming our stack is inherently lighter.

Amygdala: Prompt a topic, get a footnoted report from experts in seconds

https://amygdala.eu/research
1•JoranCornelisse•36s ago•0 comments

Microsoft EVP Rajesh Jha is retiring

https://www.businessinsider.com/microsoft-executive-rajesh-jha-retiring-read-memos-2026-3
1•DarkCrusader2•1m ago•1 comments

Military AI as 'Abnormal' Technology

https://www.lawfaremedia.org/article/military-ai-as--abnormal--technology
2•hn_acker•1m ago•0 comments

Use Neovim's server capabilities on your local machine

https://github.com/mhinz/neovim-remote
1•RobTonino•2m ago•0 comments

UK to drop historic figures from banknotes and change them to images of wildlife

https://brusselssignal.eu/2026/03/uk-to-drop-historic-figures-from-banknotes-and-change-them-to-i...
2•thunderbong•2m ago•0 comments

2026 AI Adoption and Workforce Performance Benchmarks

https://www.activtrak.com/resources/state-of-the-workplace/
1•ptrhvns•2m ago•0 comments

Runtime Safety Infrastructure for AI Agents

https://nono.sh
1•TheTaytay•3m ago•0 comments

Show HN: OneCLI – Vault for AI Agents in Rust

https://github.com/onecli/onecli
1•guyb3•4m ago•0 comments

Finland Is Ready for Russia. Is Anyone Else?

https://www.bloomberg.com/graphics/2026-opinion-finland-is-ready-russia-arctic/
1•RyanShook•5m ago•0 comments

Get ready for takeoff with Uber and Joby

https://www.uber.com/us/en/newsroom/uber-air/
1•r-bt•5m ago•0 comments

Live AI Session Summaries in a Two-Line Tmux Status Bar

https://quickchat.ai/post/tmux-session-summaries-for-parallel-ai-agents
1•piotrgrudzien•6m ago•0 comments

Reversing memory loss via gut-brain communication

https://med.stanford.edu/news/all-news/2026/03/gut-brain-cognitive-decline.html
1•mustaphah•6m ago•0 comments

Should Sam Altman fear token compression?

3•Gillesray•8m ago•0 comments

Show HN: Open-Source GTM Skills for Claude Code, Codex, and Cursor

https://github.com/athina-ai/goose-skills
1•hbamoria•8m ago•0 comments

DesiPeeps

https://desipeeps.com
1•saibuilds•10m ago•0 comments

Should Sam Altman fear token compression?

https://www.edgee.ai/blog/posts/2026-03-12-should-sam-altman-fear-token-compression-technology-or...
1•Gillesray•10m ago•1 comments

I wrote Gitleaks, now I'm maintaining Betterleaks

https://www.aikido.dev/blog/betterleaks-gitleaks-successor
4•zricethezav•11m ago•1 comments

Nvidia Fork of Godot Engine

https://github.com/NVIDIA-RTX/godot
1•throwaway2027•11m ago•0 comments

Hegger

https://hegger.party
2•davedx•12m ago•0 comments

Ask HN: Which DNS based ad blocker do you suggest?

2•SoftwareEn2•12m ago•1 comments

Save the Student Essay

https://openquestionsblog.substack.com/p/save-the-student-essay
1•voxleone•13m ago•0 comments

Show HN: BoltzPay – fetch() that pays for AI agents (x402 and L402)

https://github.com/leventilo/boltzpay
1•leventilo•14m ago•0 comments

Show HN: Stop AI Debugging with Print(). Use a Debugger

https://github.com/AlmogBaku/debug-skill
1•almogbaku•15m ago•0 comments

Show HN: Claude Status

https://github.com/gmr/claude-status
1•crad•16m ago•0 comments

AI isn't digital anymore. It's a 1-GW power problem

1•TheBottlenecker•16m ago•0 comments

Show HN: OpenTabs – Your AI calls Slack's internal API through the browser

https://github.com/opentabs-dev/opentabs
1•Jbced•17m ago•0 comments

What we learned building 100 API integrations with OpenCode

https://nango.dev/blog/learned-building-100-api-integrations-with-opencode
1•rguldener•18m ago•0 comments

Important Updates to GitHub Copilot for Students

https://github.com/orgs/community/discussions/189268
2•archb•19m ago•1 comments

We will come to regret our every use of AI

https://libresolutions.network/articles/ai-regret/
3•paulnpace•19m ago•0 comments

Show HN: Subagent-CLI – a CLI for managing multiple coding agents

1•otakumesi•20m ago•0 comments