Arcade.dev | The MCP Runtime

Making AI actually do things for you |

[Arcade.dev | The MCP Runtime @TryArcade]

Last week Arcade & @tavilyai co-hosted a private dinner in Boston for #BosTechWeek w/ 17 senior AI leaders from JPMorgan Chase, Morgan Stanley, CVS Health, and more.

The theme at every table: everyone has agents now.

Governance and observability are where everyone's stuck.

[Richard MacManus @ricmac]

"If I were building a website tomorrow, or building a SaaS app tomorrow, I would probably build it MCP-first, because that’s how agents increasingly want to consume this."

Love this quote from @TheMostlyGreat from @TryArcade, who I interviewed for AWN. https://t.co/Edkk79ekHH

[Mateo Torres @torresmateo]

🔴 going live today: bring your own identity provider to your MCP gateway

User Sources = bring-your-own-auth for MCP gateways. plug in Okta / Entra / Ping (or Auth0, Clerk, Stytch) as the auth layer. your devs sign in with the SSO they already use.

[Arcade.dev | The MCP Runtime @TryArcade]

Arcade is heading to Microsoft Build in SF.

@avoguru is hosting an expert meetup Tues 12-1:20 & Wed 4-6, demoing Arcade live.

@nbarbettini is running a table talk Tues & Wed at 1:15 on production-grade MCP. Bring the problems you're stuck on.

Come find us!

[Arcade.dev | The MCP Runtime @TryArcade]

Deploying an AI agent to 10 people is easy. 10,000 is where it gets complicated.

Arcade MCP Gateways now connect to your existing IdP.

Every tool call runs under two identities: the agent's permitted tools and the user's verified identity.

https://t.co/26g80fIJ7Z

[Arcade.dev | The MCP Runtime @TryArcade]

Our co-founder and CTO @SamPartee gave a talk at @mcpsummit on what actually makes a good tool for agents.

He's built north of 10,000 tools.

The 10x lever is not what most people think.

Watch 👇 https://t.co/Rk9ZSnx7hu

[Arcade.dev | The MCP Runtime @TryArcade]

If you are in Boston this week, come have dinner with us.

Arcade and @tavilyai are hosting a private dinner tomorrow during #BosTechWeek for senior AI leaders working through the hard parts of getting agents into production.

Apply: https://t.co/vVJtr2GmHV

[Nasdaq Exchange @NasdaqExchange]

“We make it possible to connect an AI system and agent to those business systems in a secure and safe way.”

In partnership with @Wing_VC's 2026 Enterprise Tech 30, Alex Salazar, Co-Founder & CEO of @TryArcade, visits MarketSite to address one of enterprise AI's most critical unsolved challenges that led him to found Arcade. Watch the full interview here: https://t.co/ZjtZ2xSp3b

[Arcade.dev | The MCP Runtime @TryArcade]

MCP has won. 78% of enterprise AI teams have at least one MCP-backed agent in production.

Until now, no one had measured the ecosystem underneath.

We did, and what we found matters for anyone building agents in production.

Report: https://t.co/IXZh6s3wam

[Arcade.dev | The MCP Runtime @TryArcade]

NeMo Claw didn't like its guardrail. So it replaced it.

Anthropic's research confirms it: agents choose task completion over safety when forced to pick.

Enforcement has to live outside the reasoning layer.

Our CEO @TheMostlyGreat on why the gateway is just the front door ⬇️

[Arcade.dev | The MCP Runtime @TryArcade]

The buyer is becoming the agent. That was the opening observation at our dinner this week in SF with @tavilyai during @LangChain_OSS #Interrupt26.

The conversation ended on skills and governance. Tool quality is where that work begins.

[Agentic AI Foundation @AgenticAIFdn]

Alex Salazar from @TryArcade gave an interactive talk that ended with a live demo, agents checking his real inbox through Arcade as an MCP gateway, audience along for the ride.

That is the spirit of AGNTCon + MCPCon Europe. If you are building with agents and MCP, the community wants to hear what you are working on, what is working, and what is not.

CFP closes June 8. Submit your talk here: https://t.co/ztJGJepcgb

[Arcade.dev | The MCP Runtime @TryArcade]

Watch the video breakdown from @torresmateo: https://t.co/GMYl2mB5xm

[Arcade.dev | The MCP Runtime @TryArcade]

Every AI agent that stores OAuth tokens server-side is building a token vault. The more integrations it supports, the more valuable that vault becomes.

The Vercel breach showed what happens when that vault gets breached. One intrusion, many victims.

https://t.co/hUf4ZuB6As

[Mateo Torres @torresmateo]

"MCP adopted OAuth. Google uses OAuth. so I can just pass the token through, right?"

that single very reasonable assumption is the reason URL elicitation exists.

I sat down with @nbarbettini to recap his MCP Dev Summit talk.

[Arcade.dev | The MCP Runtime @TryArcade]

Today at @LangChain_OSS #Interrupt26, Arcade was featured in the keynote as an MCP gateway partner for LangSmith Deployment.

Together, we're setting the benchmark for partnerships pushing the industry forward on open standards, neutrality & scale. 🤝 Come say hi at the show!

[Arcade.dev | The MCP Runtime @TryArcade]

Governance is the least exciting part of agents. It is also one of the most important.

5/13, SF. With @tavilyai, we’re hosting a dinner for senior AI leaders working through the hard part: building the control plane that lets agents work at scale.

Apply: https://t.co/NO2Rwp0Uhu

[Agentic AI Foundation @AgenticAIFdn]

Shub Argha from member company @TryArcade and @Dpbrinkm chatted about a killer play for Claude that is basically a cheat code for agentic AI. The star of the conversation was the Browser Use skill, which is being plugged directly into a custom-built sandbox to give Claude a secure pair of eyes on the live web. By stacking it with the Arcade skill to handle the setup and authentication they have managed to strip away every bit of traditional infrastructure friction. This is the ultimate plug and play blueprint for anyone looking to scale secure web automation within Claude without the headache.

What's your favorite Claude skill? Share in the comments below!

[Arcade.dev | The MCP Runtime @TryArcade]

Arcade sponsored a CIO & CISO Think Tank in Dallas.

The pace of AI is outrunning most enterprises' ability to govern it. That was the honest convo.

How do you give your teams the confidence to move w/ the right guardrails in place? Still the hardest question in enterprise AI.

[Arcade.dev | The MCP Runtime @TryArcade]

Watch @torresmateo walk through it: https://t.co/TXkhOof11j

[Arcade.dev | The MCP Runtime @TryArcade]

We just open-sourced Agent Library, a local-first memory layer for AI agents built by @SamPartee.

Your agent's memory should be something you can hold in your hand. A file you own, diff, and roll back.

More on the blog: https://t.co/kMcU35hzKh