Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

While still early, Stripe MCP installs are already up 3.5x and usage is up 2x since launching our OAuth-enabled server with Anthropic Claude—rather than that thing where you authenticate via API key. (I've always wondered the uplift of not manually fumbling with API keys.)

Jeff Weinstein

47,536 subscribers

53,914 Aufrufe • vor 1 Jahr •via X (Twitter)

Wissenschaft & Technologie Bildung Finanzen

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

I hooked up Logseq 🪵 to @anthropic Claude. You can chat with, search, create, and edit Logseq notes in the Claude desktop app. It works by connecting to the local Logseq api using an MCP server bridge. (Work in progress)

I hooked up Logseq 🪵 to @anthropic Claude. You can chat with, search, create, and edit Logseq notes in the Claude desktop app. It works by connecting to the local Logseq api using an MCP server bridge. (Work in progress)

Brian Sunter

18,296 Aufrufe • vor 1 Jahr

MCP is a pain in the ass. To fix that, we’re launching DAuth: the first secure, multi-tenant MCP auth solution. OAuth isn’t enough. It doesn't support dynamically passing API keys, and you can’t trust passing credentials to non-OAuth servers, since secrets are exposed. This forces existing MCP marketplaces to be OAuth-centric and single-tenant. But OAuth is annoying to implement and assumes security expertise. DAuth works for every auth case. We encrypt your credentials on the client-side, guaranteeing MCP servers never see them. You can finally ship agents with peace of mind. For the first time, MCP is ready for takeoff 🛫

MCP is a pain in the ass. To fix that, we’re launching DAuth: the first secure, multi-tenant MCP auth solution. OAuth isn’t enough. It doesn't support dynamically passing API keys, and you can’t trust passing credentials to non-OAuth servers, since secrets are exposed. This forces existing MCP marketplaces to be OAuth-centric and single-tenant. But OAuth is annoying to implement and assumes security expertise. DAuth works for every auth case. We encrypt your credentials on the client-side, guaranteeing MCP servers never see them. You can finally ship agents with peace of mind. For the first time, MCP is ready for takeoff 🛫

Dedalus Labs

1,037,030 Aufrufe • vor 7 Monaten

I'm still picking my jaw off the floor with this MCP update: Postman will now automatically build an MCP server for an API. Automatically! You select your API and the requests you want to include, and Potman will generate a Zip file with everything you need. In collaboration with Postman, I recorded a video showing how I generated an MCP server for the Google Maps API. It's glorious! Also, and before I forget: You can now test any MCP server with your Postman application. They built MCP right in, so now you can: • Connect to an MCP server • List out all of the available tools • Test any of the tools This is a 10x boost for the MCP ecosystem. If you've tried the default MCP debugger, you'll understand how huge a deal it is that you can do all of that from Postman now. Go here to give this a try:

I'm still picking my jaw off the floor with this MCP update: Postman will now automatically build an MCP server for an API. Automatically! You select your API and the requests you want to include, and Potman will generate a Zip file with everything you need. In collaboration with Postman, I recorded a video showing how I generated an MCP server for the Google Maps API. It's glorious! Also, and before I forget: You can now test any MCP server with your Postman application. They built MCP right in, so now you can: • Connect to an MCP server • List out all of the available tools • Test any of the tools This is a 10x boost for the MCP ecosystem. If you've tried the default MCP debugger, you'll understand how huge a deal it is that you can do all of that from Postman now. Go here to give this a try:

Santiago

215,882 Aufrufe • vor 1 Jahr

The MCP finally has proper authentication! It now supports OAuth 2.1 out of the box. That means proper authentication for AI agents and MCP tools is finally standardized. No more hacked-together API keys or custom token logic. With OAuth 2.1 + MCP, you can now: - Dynamically register clients - Issue & verify scoped JWT tokens - Track which agent acted for which user - Plug into existing IdPs (Auth0, Clerk, Firebase, etc.) I’ve been testing it with Scalekit, a lightweight OAuth layer built for MCP servers. It adds full discovery, token flow, and audit logs with minimal config. This is the missing piece for secure, multi-agent systems. Auth that actually works.

The MCP finally has proper authentication! It now supports OAuth 2.1 out of the box. That means proper authentication for AI agents and MCP tools is finally standardized. No more hacked-together API keys or custom token logic. With OAuth 2.1 + MCP, you can now: - Dynamically register clients - Issue & verify scoped JWT tokens - Track which agent acted for which user - Plug into existing IdPs (Auth0, Clerk, Firebase, etc.) I’ve been testing it with Scalekit, a lightweight OAuth layer built for MCP servers. It adds full discovery, token flow, and audit logs with minimal config. This is the missing piece for secure, multi-agent systems. Auth that actually works.

Arindam Majumder 𝕏

11,525 Aufrufe • vor 10 Monaten

New course: MCP: Build Rich-Context AI Apps with Anthropic. Learn to build AI apps that access tools, data, and prompts using the Model Context Protocol in this short course, created in partnership with Anthropic Anthropic and taught by Elie Schoppik Elie Schoppik, its Head of Technical Education. Connecting AI applications to external systems that bring rich context to LLM-based applications has often meant writing custom integrations for each use case. MCP is an open protocol that standardizes how LLMs access tools, data, and prompts from external sources, and simplifies how you provide context to your LLM-based applications. For example, you can provide context via third-party tools that let your LLM make API calls to search the web, access data from local docs, retrieve code from a GitHub repo, and so on. MCP, developed by Anthropic, is based on a client-server architecture that defines the communication details between an MCP client, hosted inside the AI application, and an MCP server that exposes tools, resources, and prompt templates. The server can be a subprocess launched by the client that runs locally or an independent process running remotely. In this hands-on course, you'll learn the core architecture behind MCP. You’ll create an MCP-compatible chatbot, build and deploy an MCP server, and connect the chatbot to your MCP server and other open-source servers. Here’s what you’ll do: - Understand why MCP makes AI development less fragmented and standardizes connections between AI applications and external data sources - Learn the core components of the client-server architecture of MCP and the underlying communication mechanism - Build a chatbot with custom tools for searching academic papers, and transform it into an MCP-compatible application - Build a local MCP server that exposes tools, resources, and prompt templates using FastMCP, and test it using MCP Inspector - Create an MCP client inside your chatbot to dynamically connect to your server - Connect your chatbot to reference servers built by Anthropic’s MCP team, such as filesystem, which implements filesystem operations, and fetch, which extracts contents from the web as markdown - Configure Claude Desktop to connect to your server and others, and explore how it abstracts away the low-level logic of MCP clients - Deploy your MCP server remotely and test it with the Inspector or other MCP-compatible applications - Learn about the roadmap for future MCP development, such as multi-agent architecture, MCP registry API, server discovery, authorization, and authentication MCP is an exciting and important technology that lets you build rich-context AI applications that connect to a growing ecosystem of MCP servers, with minimal integration work. Please sign up here!

New course: MCP: Build Rich-Context AI Apps with Anthropic. Learn to build AI apps that access tools, data, and prompts using the Model Context Protocol in this short course, created in partnership with Anthropic Anthropic and taught by Elie Schoppik Elie Schoppik, its Head of Technical Education. Connecting AI applications to external systems that bring rich context to LLM-based applications has often meant writing custom integrations for each use case. MCP is an open protocol that standardizes how LLMs access tools, data, and prompts from external sources, and simplifies how you provide context to your LLM-based applications. For example, you can provide context via third-party tools that let your LLM make API calls to search the web, access data from local docs, retrieve code from a GitHub repo, and so on. MCP, developed by Anthropic, is based on a client-server architecture that defines the communication details between an MCP client, hosted inside the AI application, and an MCP server that exposes tools, resources, and prompt templates. The server can be a subprocess launched by the client that runs locally or an independent process running remotely. In this hands-on course, you'll learn the core architecture behind MCP. You’ll create an MCP-compatible chatbot, build and deploy an MCP server, and connect the chatbot to your MCP server and other open-source servers. Here’s what you’ll do: - Understand why MCP makes AI development less fragmented and standardizes connections between AI applications and external data sources - Learn the core components of the client-server architecture of MCP and the underlying communication mechanism - Build a chatbot with custom tools for searching academic papers, and transform it into an MCP-compatible application - Build a local MCP server that exposes tools, resources, and prompt templates using FastMCP, and test it using MCP Inspector - Create an MCP client inside your chatbot to dynamically connect to your server - Connect your chatbot to reference servers built by Anthropic’s MCP team, such as filesystem, which implements filesystem operations, and fetch, which extracts contents from the web as markdown - Configure Claude Desktop to connect to your server and others, and explore how it abstracts away the low-level logic of MCP clients - Deploy your MCP server remotely and test it with the Inspector or other MCP-compatible applications - Learn about the roadmap for future MCP development, such as multi-agent architecture, MCP registry API, server discovery, authorization, and authentication MCP is an exciting and important technology that lets you build rich-context AI applications that connect to a growing ecosystem of MCP servers, with minimal integration work. Please sign up here!

Andrew Ng

142,137 Aufrufe • vor 1 Jahr

i hate the way tool calling works today, so i built a fix for it its called `executor` and it lets your agent run code to call any API / MCP / GraphQL server rather than installing CLIs this is a soft launch, you can try it with npm i -g executor but im still making it great

i hate the way tool calling works today, so i built a fix for it its called `executor` and it lets your agent run code to call any API / MCP / GraphQL server rather than installing CLIs this is a soft launch, you can try it with npm i -g executor but im still making it great

Rhys

130,265 Aufrufe • vor 5 Monaten

The guy who vibe-coded his own food discovery app just dropped 4 solid security basics. If you're building your own app with AI, you need this so you don't get hacked on day one: 1. Hide your API keys Keep your OpenAI or Stripe keys off the frontend to prevent them from being stolen. Tell Claude: "Move all API keys to the server and use environment variables." 2. Input Validation Stop people from pasting malicious code into your search bars to break your database. Tell Claude Code: "Sanitize all user inputs." 3. Don't invent your own auth Building login systems from scratch is too risky. Use trusted platforms instead. Tell Claude Code: "Integrate Supabase for authentication." 4. Rate limiting Prevent trolls from crashing your server with 10,000 requests per minute. Tell Claude Code: "Add rate limiting to my API routes." Too many people vibe-coding right now make these exact mistakes and wake up to drained API budgets or wiped databases. Get the security right first. Once your foundation is solid, you can build literally anything. A friend of mine just shared a crazy guide on how to build an entire social network like X in a certain niche with Claude Code in 4 weeks. check it out below 👇

The guy who vibe-coded his own food discovery app just dropped 4 solid security basics. If you're building your own app with AI, you need this so you don't get hacked on day one: 1. Hide your API keys Keep your OpenAI or Stripe keys off the frontend to prevent them from being stolen. Tell Claude: "Move all API keys to the server and use environment variables." 2. Input Validation Stop people from pasting malicious code into your search bars to break your database. Tell Claude Code: "Sanitize all user inputs." 3. Don't invent your own auth Building login systems from scratch is too risky. Use trusted platforms instead. Tell Claude Code: "Integrate Supabase for authentication." 4. Rate limiting Prevent trolls from crashing your server with 10,000 requests per minute. Tell Claude Code: "Add rate limiting to my API routes." Too many people vibe-coding right now make these exact mistakes and wake up to drained API budgets or wiped databases. Get the security right first. Once your foundation is solid, you can build literally anything. A friend of mine just shared a crazy guide on how to build an entire social network like X in a certain niche with Claude Code in 4 weeks. check it out below 👇

Spivach

11,103 Aufrufe • vor 2 Monaten

I vibe-coded an AI agent for making presentations 🤖🖼️ - accessible to everyone and open-source! You can upload both your context files, as well as a style template (a file that captures the style of the presentation you want to create). Then you launch a prompt and watch the agent one-shot your presentation. After the presentation is generated, you can edit the text inline, chat with the agent to add/modify/delete slides, and export to Powerpoint or PDF. Built with Claude Code, powered by Claude Agent SDK + LlamaParse. Hosted on Vercel/Render. To use it, you do need to specify your LlamaCloud API key and Anthropic API key (see below). App: Repo: Sign up to LlamaCloud:

I vibe-coded an AI agent for making presentations 🤖🖼️ - accessible to everyone and open-source! You can upload both your context files, as well as a style template (a file that captures the style of the presentation you want to create). Then you launch a prompt and watch the agent one-shot your presentation. After the presentation is generated, you can edit the text inline, chat with the agent to add/modify/delete slides, and export to Powerpoint or PDF. Built with Claude Code, powered by Claude Agent SDK + LlamaParse. Hosted on Vercel/Render. To use it, you do need to specify your LlamaCloud API key and Anthropic API key (see below). App: Repo: Sign up to LlamaCloud:

Jerry Liu

21,073 Aufrufe • vor 7 Monaten

If you're juggling 10+ API Keys like me, this video is gonna save your life Agent Cash has 1000s of API Keys that you can access from ONE Solana wallet with USDC instead of needing to link 10 cards You load up a wallet with USDC and whenever you need to call an API from the 1000s they have, it uses your balance from there Google Veo, Nano Banana 2, X API, Web Search, Prices and 1000s more are all accessible from one place now instead of needing to link 10 cards to 10 different places Here's me showing how easy it is to give to your Openclaw and then making a video of a flying dinosaur with it

If you're juggling 10+ API Keys like me, this video is gonna save your life Agent Cash has 1000s of API Keys that you can access from ONE Solana wallet with USDC instead of needing to link 10 cards You load up a wallet with USDC and whenever you need to call an API from the 1000s they have, it uses your balance from there Google Veo, Nano Banana 2, X API, Web Search, Prices and 1000s more are all accessible from one place now instead of needing to link 10 cards to 10 different places Here's me showing how easy it is to give to your Openclaw and then making a video of a flying dinosaur with it

ashen

38,162 Aufrufe • vor 5 Monaten

Global payroll is one of the hardest problems in business. We've spent 7 years solving it. Today we open it up, and step into our next chapter as the leading global employment infrastructure. Any company, any tool, any AI agent can now connect directly to Remote via MCP. No API keys. No custom integrations. Just the infrastructure tens of thousands of companies already run on. Watch Job explain👇

Global payroll is one of the hardest problems in business. We've spent 7 years solving it. Today we open it up, and step into our next chapter as the leading global employment infrastructure. Any company, any tool, any AI agent can now connect directly to Remote via MCP. No API keys. No custom integrations. Just the infrastructure tens of thousands of companies already run on. Watch Job explain👇

Remote

4,424,205 Aufrufe • vor 3 Monaten

You are brain-dead if you think the CLI killed MCP. Anthropic just reported 300 million downloads for the MCP SDK (up from 100M at the beginning of the year). The complaint I keep hearing: MCP pollutes your context. That's a 1-year-old argument. Context pollution is not a protocol problem, but a client one. None of the agentic tools you use have that issue: Claude Code, Codex, Cursor, etc. All of them implement progressive disclosure and load MCP tools on demand. MCPs and CLIs complement each other: • MCP offers OAuth-based authentication, standardized transport, remote access for cloud-hosted agents, and a public registry with thousands of servers. • CLIs offer Unix-style composition (pipes, stdin/stdout), zero upfront schema cost, and massive training distribution (models have seen far more shell than any structured tool format). You don't (and shouldn't) pick a side here. You should be using both where they fit.

You are brain-dead if you think the CLI killed MCP. Anthropic just reported 300 million downloads for the MCP SDK (up from 100M at the beginning of the year). The complaint I keep hearing: MCP pollutes your context. That's a 1-year-old argument. Context pollution is not a protocol problem, but a client one. None of the agentic tools you use have that issue: Claude Code, Codex, Cursor, etc. All of them implement progressive disclosure and load MCP tools on demand. MCPs and CLIs complement each other: • MCP offers OAuth-based authentication, standardized transport, remote access for cloud-hosted agents, and a public registry with thousands of servers. • CLIs offer Unix-style composition (pipes, stdin/stdout), zero upfront schema cost, and massive training distribution (models have seen far more shell than any structured tool format). You don't (and shouldn't) pick a side here. You should be using both where they fit.

Santiago

68,884 Aufrufe • vor 4 Monaten

There's no way this is free. Anthropic just put 20+ courses on using AI for $0 These are the official Anthropic courses for learning Claude and Claude Code. 20 courses that take you from zero to building real, complex things. Link: You'll find everything there: from AI basics, to advanced skills like building with MCP, the API, AWS, agents, and a lot more. And it's not just lectures. There are hands-on lessons where you practice what you learned. After each one, you get a certificate you can add to your LinkedIn, the kind that sets you apart from people who don't have it. Now you've got no excuse left for not knowing how to use AI. Bookmark this

There's no way this is free. Anthropic just put 20+ courses on using AI for $0 These are the official Anthropic courses for learning Claude and Claude Code. 20 courses that take you from zero to building real, complex things. Link: You'll find everything there: from AI basics, to advanced skills like building with MCP, the API, AWS, agents, and a lot more. And it's not just lectures. There are hands-on lessons where you practice what you learned. After each one, you get a certificate you can add to your LinkedIn, the kind that sets you apart from people who don't have it. Now you've got no excuse left for not knowing how to use AI. Bookmark this

Yarchi

15,915 Aufrufe • vor 1 Monat

This is Cortical Cloud. Live neural networks that you can interact with and train! Now open to the public. What will you discover? Credits: Frank Yang and a big thank you and acknowledgement to the rest of the Cortical Labs team. Sign up for Cortical Cloud: Learn more about us: Check out our API: Check out our API Docs: Check out our Developer Guide: Join our Discord:

This is Cortical Cloud. Live neural networks that you can interact with and train! Now open to the public. What will you discover? Credits: Frank Yang and a big thank you and acknowledgement to the rest of the Cortical Labs team. Sign up for Cortical Cloud: Learn more about us: Check out our API: Check out our API Docs: Check out our Developer Guide: Join our Discord:

Cortical Labs

227,447 Aufrufe • vor 5 Monaten

How to use 50+ API keys (models) for FREE on OpenClaw API??? - go to - login or register your account - click on "more models" - click on "use case" and select what you need it for - choose the model and open it - click on "view code" → "Generate API key" many models don't allow direct deploy, so use the "view code" button to generate API access basically Nvidia NIM gives you the ability to test almost any model from their list for FREE some of them are not worse than GPT 5.2 or Claude Opus 4.6, some might even perform better depending on the task how to understand if a model is efficient and compare it with others??? - go to - type the model name in search - click on "benchmarks" - you’ll see performance tests and rankings this way you can easily compare free models with paid ones of course there are RPM limits, on many models it’s around ~40 requests per minute each model is different, after generating the API key, RPM limits are shown in the top-right corner nothing stops you from using them, many work perfectly fine, super solid option for first tests and for learning OpenClaw or any other system where you need an AI API model

How to use 50+ API keys (models) for FREE on OpenClaw API??? - go to - login or register your account - click on "more models" - click on "use case" and select what you need it for - choose the model and open it - click on "view code" → "Generate API key" many models don't allow direct deploy, so use the "view code" button to generate API access basically Nvidia NIM gives you the ability to test almost any model from their list for FREE some of them are not worse than GPT 5.2 or Claude Opus 4.6, some might even perform better depending on the task how to understand if a model is efficient and compare it with others??? - go to - type the model name in search - click on "benchmarks" - you’ll see performance tests and rankings this way you can easily compare free models with paid ones of course there are RPM limits, on many models it’s around ~40 requests per minute each model is different, after generating the API key, RPM limits are shown in the top-right corner nothing stops you from using them, many work perfectly fine, super solid option for first tests and for learning OpenClaw or any other system where you need an AI API model

Ronin

58,940 Aufrufe • vor 6 Monaten

A lot of people asked how to connect Burp Suite MCP with OpenCode after my burpsuite mcp with Claude Code guide. So here you go 👇 1. Fire up Burp → Install the MCP extension → Run the required BMCP setup checks 2. Follow the setup guide → BMCP + Claude Code guide 3. Connect it to OpenCode opencode mcp add → Server name: burp → Server type: remote → Address: 127.0.0.1:9876 And that's it. I tested it by opening example./com through Burp and interacting with it using OpenCode + Burp MCP. If you're already using OpenCode for security work, this is a pretty clean way to bring Burp into the workflow. 🔥

A lot of people asked how to connect Burp Suite MCP with OpenCode after my burpsuite mcp with Claude Code guide. So here you go 👇 1. Fire up Burp → Install the MCP extension → Run the required BMCP setup checks 2. Follow the setup guide → BMCP + Claude Code guide 3. Connect it to OpenCode opencode mcp add → Server name: burp → Server type: remote → Address: 127.0.0.1:9876 And that's it. I tested it by opening example./com through Burp and interacting with it using OpenCode + Burp MCP. If you're already using OpenCode for security work, this is a pretty clean way to bring Burp into the workflow. 🔥

Abhishek

14,940 Aufrufe • vor 19 Tagen

P-Video-Avatar is now the fastest model for top-quality and diverse avatar generation. - It costs 0.025$ per second of video, making it up 6× cheaper than similar avatar tools - It takes 1.83s generation time per second of video, making it up to 18x faster than similar avatar tools. - It offers full-body control, integrated audio support in 20+ languages, and is long-form video-ready for up to 3-minute consecutive clips. We are launching with free usage until Sunday 23:59 CET at Replicate, Wiro AI — Ship AI Faster, inference shell, Runware, Scenario And with general availability at: DeepInfra, each::labs Try it now: 📚 API & quickstart for Pruna API: ⭐️ Star our OSS: 🌐 Website:

Pruna AI

14,855 Aufrufe • vor 4 Monaten

TRUMP: “Inflation is down. All costs are down. Everything is down, other than the thing that you carry the babies around in. Energy is down. Gasoline is down. Far ahead of schedule. So people that were paying a lot of groceries, they are paying a lot less.” FACT Check: - Gasoline is up since Trump took office (from $3.13 to $3.17) - Grocery prices, on average, are up since he took office. - Inflation is still here, still above where it should be. - Energy prices are down mostly because of recession fears. - All costs are definitely not down. Most are up.

TRUMP: “Inflation is down. All costs are down. Everything is down, other than the thing that you carry the babies around in. Energy is down. Gasoline is down. Far ahead of schedule. So people that were paying a lot of groceries, they are paying a lot less.” FACT Check: - Gasoline is up since Trump took office (from $3.13 to $3.17) - Grocery prices, on average, are up since he took office. - Inflation is still here, still above where it should be. - Energy prices are down mostly because of recession fears. - All costs are definitely not down. Most are up.

Ed Krassenstein

578,197 Aufrufe • vor 1 Jahr

Anthropic won't like this open-source repo. It is going to cost LLM providers a lot of money. Every CI run of an AI app today sends real requests to providers like OpenAI or Anthropic. Like any other LLM call, this too gets billed at actual API rates. So for teams with high commit volumes, this accumulates into a meaningful chunk of API spend. One common hack devs use is that instead of invoking the LLM API, the test calls a fake local server that speaks the same API and returns a dummy response. The catch is that the dummy response is a copy of what the provider returned on the day it was saved, and providers keep adding fields and changing types. So the tests keep passing against a schema that's no longer valid, while the real integration breaks in production. A smart approach is now actually implemented in CopilotKit🪁's recently open-sourced aimock project. Every day, the repo's own CI sends a handful of requests to the real API and the same requests to the fake server, then compares both against the official client library's type definitions. Those are the only real API calls in the whole setup, and they run on the repo's own keys, not in anyone else's CI. A single team can push hundreds of commits a day, and thousands of teams are already doing that with coding agents. All of those runs stay offline, because one repo checks against the real API on everyone's behalf. When a check fails, a coding agent updates aimock's built-in response schema, the full test suite has to pass, and a patch version ships to npm. By simply upgrading the package, the corrected schema gets reflected in every project using it. The capability is not just limited to a single provider. The same server works for Claude, OpenAI, Gemini, Bedrock, Azure, Ollama, plus MCP tools, A2A agents, AG-UI event streams, vector DBs like Pinecone and Qdrant, and search, speech, image, and video endpoints. Here's the repo: (don't forget to star it ⭐) That said, mocking your API calls is one thing. AI engineers should also know how to test agents properly in the first place, which several teams still skip. I wrote a full walkthrough on that, covering build, testing, evals, tracing, and deployment. Read it below.

Anthropic won't like this open-source repo. It is going to cost LLM providers a lot of money. Every CI run of an AI app today sends real requests to providers like OpenAI or Anthropic. Like any other LLM call, this too gets billed at actual API rates. So for teams with high commit volumes, this accumulates into a meaningful chunk of API spend. One common hack devs use is that instead of invoking the LLM API, the test calls a fake local server that speaks the same API and returns a dummy response. The catch is that the dummy response is a copy of what the provider returned on the day it was saved, and providers keep adding fields and changing types. So the tests keep passing against a schema that's no longer valid, while the real integration breaks in production. A smart approach is now actually implemented in CopilotKit🪁's recently open-sourced aimock project. Every day, the repo's own CI sends a handful of requests to the real API and the same requests to the fake server, then compares both against the official client library's type definitions. Those are the only real API calls in the whole setup, and they run on the repo's own keys, not in anyone else's CI. A single team can push hundreds of commits a day, and thousands of teams are already doing that with coding agents. All of those runs stay offline, because one repo checks against the real API on everyone's behalf. When a check fails, a coding agent updates aimock's built-in response schema, the full test suite has to pass, and a patch version ships to npm. By simply upgrading the package, the corrected schema gets reflected in every project using it. The capability is not just limited to a single provider. The same server works for Claude, OpenAI, Gemini, Bedrock, Azure, Ollama, plus MCP tools, A2A agents, AG-UI event streams, vector DBs like Pinecone and Qdrant, and search, speech, image, and video endpoints. Here's the repo: (don't forget to star it ⭐) That said, mocking your API calls is one thing. AI engineers should also know how to test agents properly in the first place, which several teams still skip. I wrote a full walkthrough on that, covering build, testing, evals, tracing, and deployment. Read it below.

Akshay 🚀

62,821 Aufrufe • vor 18 Tagen