MCP App Store
by KeenEthics
Simmie icon

Simmie MCP Server for Claude and ChatGPT

By Simmie

Overview

Simmie is a sales training platform where reps practice live roleplay calls with realistic AI buyers and get scored against their team's own rubric. This connector lets managers and enablement teams set up that training from Claude. Build practice. Describe the conversation your team needs to get better at, such as cold-call objection handling or a renewal call with a churn-risk customer. Simmie creates the training module, the buyer persona and the scoring rubric, and returns a practice link your reps open in the Simmie app. Score real calls. Paste a transcript with the speakers labeled and name the rep. Simmie grades it against your organization's default rubric and saves the result to that rep's record. It can then build a practice scenario aimed at what went badly and assign it to that rep. Set your standard. Adopt one of your existing rubrics as the default for scoring calls, or describe what a good call looks like and Simmie builds one. The connector returns text and links only. Voice practice happens in the Simmie app, not in Claude. You need a Simmie account. On team accounts, creating simulations, scoring calls and changing the default rubric are limited to org and account admins.

How to connect

  1. Claude: Open the Simmie listing in the Claude connectors directory and choose Connect. Claude listing Simmie exposes 7 tools here.
  2. ChatGPT: Open the Simmie listing in the ChatGPT app directory and choose Connect. ChatGPT listing Simmie exposes 7 tools here.

Tools

create_simulation

ChatGPT
Create a complete Simmie roleplay simulation from a plain-English brief and return a shareable practice link: a training module, a persona and a rubric, saved and bound in one call. Use it whenever the user wants to make a roleplay, build a practice scenario, or create a simulation for their team. To build practice from a call that was already SCORED, use recommend_training instead: it targets that rep's real weaknesses and assigns it to them. On a team account only an org or account admin may create simulations. Always generates a NEW persona, so check list_personalities first if they may already have a suitable one. Ownership follows the account: team accounts share with the org, individual accounts stay private. Takes 45-60s and has no dedup, so wait rather than retrying.

list_personalities

ChatGPT
List the roleplay personalities (personas) the current user/org already has in Simmie. Call this before creating a simulation so you can tell the user what they already own and let them decide, or to show them their library. You cannot reuse one — create_simulation always generates a fresh persona. Read-only; takes no arguments. Returns id, name, a short summary, profession, gender, difficulty, and owner scope for each. Returns at most 50 and cannot page, so treat a missing match as 'not in the most recent 50' rather than proof the org does not have one.

list_team_members

ChatGPT
List the people in the caller's organization that a scored call can be attributed to, with their emails. Call this BEFORE score_transcript on a team account, so the user picks a real member from the actual list instead of typing a name from memory: score_transcript writes permanently to whoever is named and nothing can move it afterwards. Pass the name or email they choose as score_transcript's for_user. Read-only. Individual accounts have no team and get an empty list.

ping

ChatGPT
Liveness and wiring check. Call this once when configuring a new MCP client to confirm Simmie is reachable and the credentials are valid. Returns a static payload; takes no arguments. Use this only for setup or troubleshooting — it has no product effect.

recommend_training

ChatGPT
Build a practice simulation targeting whatever a scored call went badly on and, on a team account, assign it to that rep. Call this ONLY after score_transcript and ONLY when the user said yes to the follow-up it offered. Pass the conversation_id from the scoring result; the evidence comes from the saved call. If nothing has been scored here, do not call it: ask for the transcript, run score_transcript, then offer this. For a simulation not based on a real call use create_simulation. On a team account only an org or account admin may call this. Takes up to two minutes; if it times out, calling again with the same conversation_id is safe and re-attaches to the same job.

score_transcript

ChatGPT
Score a real sales call transcript against the organization's default rubric and PERMANENTLY save it as a scored call on a named person's record. Needs a SPEAKER-IDENTIFIED transcript: turns split by who was speaking, plus which speaker is the rep being scored. A raw unlabelled paste cannot be graded, and naming the wrong speaker silently grades the other party instead. On a team account only an org or account admin may use this, and must name the member it belongs to: call list_team_members first and let them pick. Nothing undoes this write, so confirm the speaker and the person first and stop if either is unclear. Needs a default scoring rubric on the org; if none is set this refuses and setup_scoring_rubric creates one. Returns a next_step suggesting follow-up practice.

setup_scoring_rubric

ChatGPT
Set the organization's default rubric for scoring call transcripts, either by adopting one it already has or by creating a new one. Pass exactly one of existing_rubric_title (to adopt) or criteria (a plain-English description of what a good call looks like, to build a new one); if both are given the title wins. Nothing lists rubrics for you, so only adopt a title the USER supplied, never one you inferred. On a team account only an org or account admin may call this. If a default is already set this REPLACES it silently: the tool cannot report or restore the one it displaces, and the change applies org-wide to every transcript scored from then on, so ask whether a standard is already in place first. Does not affect practice sessions, which use the rubric attached to their own module.

create_simulation

Claude

list_personalities

Claude

list_team_members

Claude

ping

Claude

recommend_training

Claude

score_transcript

Claude

setup_scoring_rubric

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/simmie"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/simmie)