28
votes
Actually useful MCPs
I'm a web developer and find the playwright MCP to be genuinely useful. My LLM is able to navigate my site, measure the size of elements, see console errors, network requests, etc. This is the only MCP I've ever installed and haven't yet had any cause to use others. But I'm interested in hearing what other professionals are using.
Not for development but I integrate with Hevy to let an LLM manage my gym workouts. It monitors my progressive overload progression, plans around my other sports, if I miss a workout it’ll redistribute my exercises, things like that.
Would love to know more. Can you explain your setup in detail? Does it automatically update your workout routines as you go? Does it do a good job of maintaining muscle groups and without variety? Have you ever noticed it didn't do a good job and had to manually change things?
Basically I have it hooked up to openclaw and obsidian, through a skill that calls Hevy MCP endpoints and the standard obsidian skill. I have five different workouts in Hevy labelled Monday-Friday. I have an obsidian document that contains my overall goals, lists of acceptable exercises. I’m integrated with telegram so every morning I get a message telling me what my workout is going to be, which is really just for fun, then I go to the gym, do the workout, sometimes leaving notes like “very easy, should increase weight”, “skipped today because rushed for time”, etc. It’s set up to poll until my workout is done and then it analyzes, proposes changes (increase reps to 12, increase weight to 60 and drop reps to 8, schedule missed exercise for Friday, etc).
Useful stuff it’s done for me lately include reorganizing my schedule around an injury and my physiotherapist’s prescribed rehab exercises, and planning out a phased return to the gym after being out for two weeks with a tonsil infection.
On occasion I’ll find that it’s doing something like scheduling too much or not enough and I’ll have to chat a bit about it but generally it’s well behaved. I maintain my obsidian vault in git and it has a copy of all my schedule changes just in case but I haven’t needed that backup yet.
That's awesome! This gives me enough ideas to implement something similar. I use Hermes Agent instead of OpenClaw and I'll set something up. Thank you for explaining this in detail. Appreciate it.
Would also recommend Liftosaur, it has a custom scripting language for workouts so you can really customize your routine however you like. It's also open source! (though nor easily selfhostable afaik)
This appears to be the source code:
https://github.com/astashov/liftosaur
By the way, note that most of the times you don't need an MCP at all, and the LLMs generally perform better when just given access to a command line script and a skill to run it. In fact, OpenClaw actually does that, it just uses a wrapper that automatically converts MCPs to standalone command line tools.
Here's the post from the developer of the Pi agent, which is also what OpenClaw is based on, specifically using the example of Playwright:
https://mariozechner.at/posts/2025-11-02-what-if-you-dont-need-mcp/
Another one not for development but I've got a friend who's been experimenting with MCPs in their day-trading routine. They use Interactive Brokers as their brokerage (who have an MCP) and also use a trading information/charting platform (unsure of name, apologies!) that also includes an MCP. It's still currently only paper trading but they've been seeing some decent success with it, having made a few thousand dollars over the course of the month of May (though all paper money, not real money). They were telling me that they've still got few months before they unleash it into the world, mainly because the paper trading system assumes any orders put into it are immediately filled when in reality, there's order queues that could potentially lead to orders not being filled. It's an interesting use of MCPs and AI that I haven't seen discussed much online.
My company’s data warehouse is in Databricks, and I use their MCP daily. Super handy to just cut Claude loose on whatever fresh hell of metric requests I have waiting for me when I log on and have a fancy looking dashboard to placate leadership within the hour.
Slack MCP is handy for me too, but I mostly use it for trends and behavior analysis in channels to supplement the metrics in the dashboard above.
I don’t think they even look at the reports :(
One thing I use often is https://github.com/BeehiveInnovations/pal-mcp-server
It’s nice to let Claude ask Gemini for a second opinion during plan mode, or when I’m translating my tool into a language I don’t speak
I'm not using any yet. I might use the Playwright MCP except that on exe.dev, the agent already has Chrome integration.
Ah, but what about Firefox integration?
Uh, I haven't tested with it. But I'm just working on my own personal websites and I'm unlikely to do anything fancy that would break Firefox. I should check that sometime.
For work I find the Slack MCP server pretty useful. I'm juggling a lot of context across many channels, and using Claude to search slack for a particular tidbit I remember someone saying a month ago is usually faster than me trying Slack search. I use Claude to manage my daily work notes and todos, and I like telling it to just read what I wrote to x person and generate notes from that rather than re-type it all into my notes manually.
For anyone out there using Linear for project management, me and my team have found their MCP integration incredibly useful. When tasks are small and well defined, Claude can generally do a B+ implementation with no other context than “read this linear issue and get to work.” Of course, YMMV, and I’d imagine it depends a lot on how heavily your team uses Linear day-to-day.
Fastmail has a very good mcp server that I used to have Hermes go through my crazy long email inbox. I also had it build out a bunch of calendar events from a pdf of an itinerary. I love their stance on ai: it's your data, so you should be able to access it in the way you want to.