In partnership with
MCP has a canonical reference repo with real weight behind it: 90,000-plus stars on the official servers repo. That matters less as a trophy and more as a signpost. When the protocol's own maintainers publish the definitive starting point, the improvisation phase is ending. If you've been hand-rolling MCP connections or copying half-documented examples from someone's gist, today's the day to go check the source instead.
2 readers pay five dollars a month so the daily stays free for everyone else. If it has earned a spot in your morning, be number 3.
|
Sponsored
|
How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads
For its first CTV campaign, Jennifer Aniston’s DTC haircare brand LolaVie had a few non-negotiables. The campaign had to be simple. It had to demonstrate measurable impact. And it had to be full-funnel.
LolaVie used Roku Ads Manager to test and optimize creatives — reaching millions of potential customers at all stages of their purchase journeys. Roku Ads Manager helped the brand convey LolaVie’s playful voice while helping drive omnichannel sales across both ecommerce and retail touchpoints.
The campaign included an Action Ad overlay that let viewers shop directly from their TVs by clicking OK on their Roku remote. This guided them to the website to buy LolaVie products.
Discover how Roku Ads Manager helped LolaVie drive big sales and customer growth with self-serve TV ads.
The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.
The Drops
Repomodelcontextprotocol/servers
90,533 stars · modelcontextprotocol/servers
The official reference list of MCP servers, maintained by the people who wrote the spec.
Here is the real problem it solves. Half the MCP tutorials online point to abandoned forks or servers that predate a breaking spec change, and you don't find out until your agent silently fails a tool call.
Quick start: clone it, skim the seven reference servers (filesystem, Git, fetch, memory, time and two more), and copy the config block instead of writing your own from scratch.
Catch: it's a reference list, not a package manager. You still wire each server in by hand.
AffiliateDify
Once the official servers are wired, the next question is where those tools actually run. One answer speaks MCP in both directions: it imports tools from an MCP server over HTTP and can publish your own app back out as one. Dify, an open-source platform for building LLM apps: visual workflows, RAG pipelines, agents, and observability in one place. Self-host it or run the cloud, and take a prototype to production without stitching a framework together yourself.
Use it for: self-hosting visual agent workflows with RAG and observability in one stack.
We may earn a commission.
ToolLLM by Simon Willison
12,537 stars · simonw/llm
A CLI and Python library that runs prompts against any model without locking you into one vendor's SDK.
Use it for: swapping between Claude, GPT, and a local model with one command when you're benchmarking cost against quality.
Catch: the plugin ecosystem is wide but uneven. Some model plugins lag behind their provider's latest release.
Repotrycua/cua
25,632 stars · trycua/cua
Open-source computer-use drivers for running agents across real desktop environments, not sandboxed demos.
Use it for: training or evaluating an agent that has to click, type, and navigate actual software, not just call APIs.
Catch: cross-OS fleets means real infrastructure to stand up. This isn't a five-minute local test.
Repoccseva
807 stars · Iamshankhadeep/ccseva
A macOS menu bar app that tracks your Claude Code usage in real time.
Use it for: catching a runaway session before it eats your weekly limit, instead of finding out at the invoice.
Catch: macOS only, and it reads local usage data, so it won't help if you're running Claude Code across a shared team account.
|
From Our Partners
|
Your identity deserves 24/7 protection
Identity theft can happen to anyone. Coveron monitors your credit, dark web, and financial activity to catch fraud before it costs you. One scam can cost you everything, protect yourself now, the first 100 users get 20% off with code beehiivenewsletter.
30-day money-back guarantee. Terms and conditions apply.
Start Here
Disconnect one thing you connected last month, and see what breaks.
1. Open the settings or integrations page for whichever AI tool you use most (ChatGPT, Claude, a browser extension, anything you gave account access to).
2. Look at the list of connected apps: your email, calendar, Google Drive, a calendar app, anything you clicked "allow" on without reading closely.
3. Pick the one you're least sure you still need.
4. Revoke its access.
5. Use the tool normally for a day. If nothing breaks, you just found a permission you didn't need to grant in the first place.
TRY THIS: Open that settings page right now and just look at the list. You don't have to disconnect anything yet, just see what you've actually given access to.
|
Recommended
Once you've cleared out the connections you don't need, the next move is filling the empty space with prompts that actually earn a place in your workflow. HubSpot's 100+ ChatGPT Prompts guide is a free download of working prompts for writing, research, and daily tasks, organized so you can lift what you need and go. We may earn a commission. |
Frontier Signals
AX, an agent execution control plane born inside Google, is now public. Its site describes an open, declarative runtime for running fleets of agents, with isolation, fast suspend and resume, and scheduling treated as core primitives. If you have been keeping idle sandboxes alive while agents wait on a model, this is the layer it targets. (agentexecutor.io)
A new read-only tool called Geiger inventories every agent, MCP server, plugin, and AI extension installed on a machine with one command. If you've lost track of what has filesystem or network access on your own laptop, this is the audit you've been putting off. (GitHub)
Vercel Connect now supports Microsoft Teams, giving any app or agent built on Vercel a managed bot that people can @mention directly in a channel. That's one less custom webhook integration for teams already living in Teams. (Vercel Blog)
Simon Willison surfaced a blunt account from an engineer two weeks into a big-company job: specs, tickets, tests, and reports are all generated by Claude Code, and "nobody knows anything" as a result. Worth sitting with if your team is scaling AI-assisted documentation faster than anyone is reading it. (Simon Willison)
Stratechery raised a sharper question about frontier labs pacing their own releases: he grants the safety motive, then argues a slowdown would also conveniently shrink the gap between what the models can do and what the labs' products and customers are ready for. (Stratechery)
|
Recommended reading
If you like The AIgent, a small group of operator-tier publications worth your inbox: see the shortlist. |
Which best describes you right now?
Before You Go
Reply and tell me which Drop you're actually going to open today. The MCP reference list is the kind of thing that's easy to bookmark and never touch, so I want to know if it's more than that for you.
See you Wednesday.
Before you go: we started a room for people actually building with agents. Tell us what you want us to build next. Join the community →
|
Want to reach builders shipping with AI every weekday? Advertise in The AIgent. |
