Tools / RateMyPrompts MCP · Coming soon
Grade your prompt set without leaving your AI.
The RateMyPrompts MCP connects Claude, ChatGPT, Cursor and other assistants to our grader. Ask for the score, the gaps and the fixes in the chat you already work in, and let the rubric check what your AI writes.
Built for
- Claude
- ChatGPT
- Cursor
- VS Code
- Windsurf
- Gemini CLI
Here's our Peec export. Grade it and tell me the three changes worth the most points.
Your set scores 58/100, Mixed. 22% of the prompts name your own brand, and two journey stages have no prompts at all.
- Trim 14 branded prompts +6
- Drop 9 near-duplicates +3
- Add real constraints to 20 prompts +2
Want me to apply them and re-score?
What it is
RateMyPrompts, as a tool your AI can use.
MCP, the Model Context Protocol, is an open standard for connecting AI assistants to other tools; its own documentation compares it to a USB-C port for AI applications. Connect the RateMyPrompts MCP once and your assistant can grade a prompt set, find what it is missing, apply fixes and plan the measurement, using the same rubric as this website.
It will not replace your tracker. Trackers tell you what assistants said about you. RateMyPrompts checks whether you asked them the right questions.
Ask it things like
Plain requests, graded answers.
“Here's our Peec export. Grade it and tell me the three changes worth the most points.”
grade_prompt_set · get_recommendations
“I've pasted ten discovery-call transcripts. Write prompts in our buyers' words, then check them before I add them.”
check_prompts
“Which parts of the buying journey is our set missing? Draft five prompts for each gap.”
find_gaps
“Remove the branded and duplicate prompts, re-score, and give me a CSV for Profound.”
apply_fixes · export_set
“We track 80 prompts daily on ChatGPT and Perplexity. Can we trust a 5-point rise?”
plan_measurement
Every answer links to a full scorecard, so the numbers can be checked, shared and reported.
How the score works →
What it will do
Eight tools, one rubric.
The first version, planned. Tell us what else you need when you join.
- Grade a prompt set
grade_prompt_set - Score a set out of 100 against the published rubric, with the seven dimensions, red flags and a link to the full scorecard.
- Check new prompts
check_prompts - Run draft prompts through the same checks before they go into your tracker: branded, leading, near-duplicate, keyword-shaped or missing a constraint.
- Find coverage gaps
find_gaps - List the topics and journey stages the set does not reach, as briefs your AI can write new prompts for.
- Get the fixes
get_recommendations - The ranked changes for a scorecard, each with the points it is worth, worked out by re-running the scorer.
- Apply fixes and re-score
apply_fixes - Apply the changes you choose, grade the revised set and return the new score next to the old one.
- Plan the measurement
plan_measurement - How many prompts, runs and assistants a set needs before a change is real, from the sample-size calculator.
- Compare versions
compare_scorecards - What changed between two versions of a set, dimension by dimension.
- Export for your tracker
export_set - The cleaned, revised set as a CSV shaped for Peec, Profound, Semrush or Ahrefs.
The loop it closes
From first draft to a set you can trust, in one conversation.
- 01
Draft
Your AI writes prompts from calls, tickets or a brief.
- 02
Check
Each prompt is checked before it goes in.
- 03
Grade
The set is scored against the published rubric.
- 04
Fix
Apply the changes worth the most points, then re-score.
- 05
Plan
Size the set to the change you need to detect.
- 06
Export
Load the clean set into your tracker.
Why connect it
The benefits of running RateMyPrompts inside your AI
Stay in the chat you already use
No exports, uploads or tab-switching. Ask for a grade in the same conversation where you are writing the prompts.
Your AI writes, the rubric checks
Assistants are good at drafting prompts and poor at judging them. The score comes from the same deterministic rubric as the website, with no model in the grade, so the AI cannot mark its own homework.
A tighter loop
Draft, grade, fix and re-score in one conversation, with the points each change is worth, instead of a round trip through a dashboard.
One method across the team
Everyone, and every client set, is measured against the same versioned rubric, so scores are comparable across people and months.
Honest numbers on demand
Ask whether a change is real before you report it. The measurement planner is one question away.
Private by default
Scorecards belong to your RateMyPrompts account and stay private. Prompt text is never shown to anyone you share a score with.
Works alongside your tracker
Trackers report the answers. This checks the questions.
AI visibility trackers such as Peec AI, Profound and SE Ranking already bring their data into assistants over MCP: how often you were mentioned, where and against whom. None of them grades the prompt set that data comes from. Connect both, and your assistant can tell you what the numbers say and whether they can be trusted.
Tracker MCPs
- Mentions and share of voice
- Citations and sources
- Trends over time
RateMyPrompts MCP
- Are these buyers' questions?
- What is missing or inflated?
- Is the change real?
How it will work
Nothing to install.
- 01
Add the connector
Paste the RateMyPrompts MCP address into your assistant's connector settings.
- 02
Sign in
Approve access with your RateMyPrompts account. Your assistant never sees a password, and you can disconnect at any time.
- 03
Ask
Paste a set or an export and ask. Results link to a private scorecard in your account.
Plans. Planned to work like the website: the same free allowance of full scorecards per verified email, and unlimited with Pro. Pricing is confirmed before anyone is charged.
What happens after you join
Early access goes to the waitlist first.
- 01
Confirm your email
One click, so nobody can sign you up without asking.
- 02
See your place
Your queue position and a personal invite link.
- 03
Move up
Every colleague who joins with your link and confirms moves you up 10 places.
- 04
Get your invite
Early access goes out in queue order. We email you once, with the link to connect.
Join the waitlist
One email field. Confirm, then see your place.
Questions
About the MCP
Want it sooner? The grader and the sample-size calculator are free on the website today.
- What is an MCP?
- MCP, the Model Context Protocol, is an open standard for connecting AI assistants to other tools. Connect the RateMyPrompts MCP once and your assistant can grade and fix prompt sets for you, without you copying anything between tabs.
- Which AI apps will it work with?
- Any assistant that supports MCP servers. We are building it for Claude, ChatGPT, Cursor, VS Code, Windsurf and Gemini CLI first. Tell us which you use when you join, and we will prioritise accordingly.
- When will it launch?
- It is being built now. Early access goes to the waitlist first, in queue order, and every confirmed referral moves you up 10 places. We will email you once when your invite is ready.
- Will it cost anything?
- It is planned to work like the website: the same free allowance of full scorecards per verified email, and unlimited with Pro. We will confirm pricing before anyone is charged.
- Do I need to be technical?
- No. In most assistants you paste one URL into the connectors settings and sign in with your RateMyPrompts account. There is nothing to install.
- Is my prompt data safe?
- Scorecards will belong to your account and stay private, as they do on the website, and the connection uses the standard sign-in flow for MCP, so your assistant never sees a password. You can disconnect at any time.
- Can I use RateMyPrompts now?
- Yes. The grader and the sample-size calculator are free on the website today. The MCP brings the same analysis into your assistant.