Ask your AI assistant about CrossFit workouts

PRzilla's workout catalog is now available to AI assistants through MCP. You can ask for a workout's prescription, benchmark targets, scaling notes, or movement details without leaving your conversation (no PRzilla account needed).

MCP, or Model Context Protocol, is a standard that lets an assistant call tools supplied by another service. In this case, those tools look things up in PRzilla's public CrossFit and HYROX catalog -- the workout you're asking about, its movements, and the available targets. We host the server, so there's nothing to run on your laptop.

Start with a workout it might not know

Try this after connecting PRzilla:

Use PRzilla to look up TJ Lobraico. Show me the full workout and its Intermediate targets for men and women, with a link to the workout.

TJ Lobraico starts with a 1-mile run, followed by 13 rounds of 9 pull-ups, 9 push-ups, 9 V-ups, 9 burpees and 9 Iron Mikes, then another 1-mile run (scored by total time).

For a "good time", the connector returns PRzilla's available performance levels along with the prescription. For this workout, the "Intermediate" (L5) targets are 58:00 for men and 1:06:45 for women. An assistant can retrieve those exact values rather than inventing a plausible range.

Those are PRzilla's estimated reference targets. The connector doesn't know your training history, so it can't say how you personally will finish. You should still check an assistant's explanation against the linked workout -- getting the data is only part of getting the answer right.

You can use familiar workouts too. Perhaps you're doing Murph and want its scaling notes, or its targets with and without a vest. That comparison is fairly straightforward for the connector (the catalog returns the available vest variants separately).

A few things to ask

There are three tools behind the conversation:

• Find workouts by name or movement, optionally narrowed to a category (e.g. Hero).
• Read a workout for its full prescription, available scaling notes and benchmark targets.
• Look up a movement for technique, equipment, easier versions, progressions and a demonstration video when one is available.

You can ask:

Use PRzilla to get Murph's scaling notes. Which options does the catalog suggest for a beginner?

Or:

Look up the thruster in PRzilla. Explain the movement and include its demonstration video.

Search results are short summaries. For a workout you plan to do, ask the assistant to fetch the full entry before judging whether it matches your equipment or movements. Some of our testing caught assistants treating a partial movement list as the whole workout (unfortunately, the answer can still sound quite convincing).

Connect it

Following Claude's setup guide, open Customize > Connectors, click Add, and choose Custom > Web. Name it "PRzilla" and enter:

https://przilla.app/mcp

Continue through the setup and choose "No sign in" for authentication -- no PRzilla login, API key or OAuth credentials are needed. Add the connector, make its tools available in your conversation, and try one of the prompts above. Claude currently allows one custom connector on its free plan -- its own plan and workspace limits still apply.

For another client that supports remote MCP over Streamable HTTP, use the same address and select no authentication if asked. The exact setup depends on the client. PRzilla also has a public Glama listing where you can inspect the three tools.

What it can see

This connection reads the public catalog. It cannot see your personal workout history, retrieve your last Fran score, or log a new result. It also cannot create a training plan in your PRzilla account.

That scope is why you don't have to sign into PRzilla. Your assistant sends a catalog lookup, and PRzilla returns the data and workout links. Our privacy policy explains what we receive and record. Your AI provider's own privacy policy still governs your conversation with it.

Try a workout you know well first, then one you don't. If a prescription or target looks off, send us the workout name and what the assistant returned at hello@przilla.app.