MCP App Store
by KeenEthics
Replio icon

Replio MCP Server for ChatGPT

By AitekLabs

Overview

Replio is an AI voice training platform: trainees practice real conversations with lifelike AI personas in live voice sessions and receive scored feedback, while managers follow their progress. This plugin lets you set up and run a Replio training project without leaving ChatGPT. Connect with your Replio account, or with an access code shared by a Replio manager, then: - Name and describe your training project - Add company knowledge: paste text and links, or upload documents through a secure page - Get training goal suggestions generated from your documents and adjust them - Generate AI training personas, then review and edit each one before going live - Build a knowledge quiz and pick the questions you want to keep - Activate the project and invite trainees - Check results afterwards: session KPIs and recent team activity Voice training sessions themselves take place on replio.eu, in 58 languages. Built in Europe, GDPR compliant, with EU data hosting.

How to connect

  1. ChatGPT: Open the Replio listing in the ChatGPT app directory and choose Connect. ChatGPT listing Replio exposes 71 tools here.

Tools

activate_project

ChatGPT
Accept the draft personas and make the project live for trainees. Omitting acceptedPersonaIds accepts every draft persona; difficultyLevels optionally overrides per-persona difficulty (2=easy, 3=medium, 4=hard). Activation cannot be undone from this connection. Writes only to this project's data in Replio; no external system is contacted.

add_contacts

ChatGPT
Add people to the organization's contact library without inviting them anywhere: contacts is a list of {email, name?, phone?} (max 500 per call), locale sets their preferred language, groupId files them into an existing group. No email is sent and no seat is used. An email already in the library is updated instead of duplicated, and a deactivated contact with that email is reactivated. Up to 50 rows run immediately; larger lists run as a background import job and the result carries a jobId. Returns created, updated and skipped (invalid or duplicate rows) counts.

add_knowledge_text

ChatGPT
Add text typed or pasted by the user (product sheet, process, team notes...) to a project's knowledge base. Replio's own pipeline extracts and indexes the content as given. Processing is asynchronous; the setup status reports when the knowledge is ready.

add_knowledge_url

ChatGPT
Fetch a public web page into the project's knowledge base: Replio's server requests the URL once and extracts its text. Processing is asynchronous; the setup status reports when the knowledge is ready.

add_library_document

ChatGPT
Creates a document in the organization's Knowledge Library from pasted text (kind text: title and text, 10 characters minimum) or from a public web page (kind url: url, optional title). The document lands in folderId, or in the library's first root folder when omitted; it is not linked to any project. Extraction runs in the background and counts as one AI operation. Identical content already in the library is returned as a duplicate instead of a new document.

add_persona

ChatGPT
Create a new live persona on an active project from the given fields, with no AI generation: the wording is stored as-is. Voice and language follow the project unless configured later in the dashboard. Refused on projects still in setup. Writes only to this project's data in Replio; no external system is contacted.

assign_contacts_to_project

ChatGPT
Enrol existing contacts (by contactIds, by groupIds, or both) in an active project and invite them. Without confirm:true the tool only returns a preview: how many people are selected, how many are already enrolled, seats used before and after, and which emails go out. With confirm:true Replio sends a localized invitation email to each newly enrolled person and uses one seat per new person. channel chooses how they practice: app (default), both (app plus phone training calls, when the project has phone training on) or phone (phone calls only, no app invitation). Max 500 selected contacts per call; deactivated contacts are skipped.

build_live_brief

ChatGPT
Force a synchronous rebuild of the brief of one Replio Live context from its linked documents (tens of seconds). Hidden cards, edits and manual cards are carried over; the objection approval is reset; the new brief is published automatically when a card was already published. Returns built, briefId and the resulting brief status. Normally not needed: linking documents or changing the language rebuilds in the background.

create_project

ChatGPT
Create a new draft training project in an organization. The project starts with no knowledge, no personas and no quiz, and is not visible to trainees until activated.

deactivate_contacts

ChatGPT
Deactivate 1 to 500 contacts of the organization. Without confirm:true the tool only returns a preview of what will happen; with confirm:true each contact is marked deactivated, every active project membership of that person in this organization is deactivated (they lose access to their trainings), and the contact is purged from the library 90 days later. Contacts already deactivated are skipped. Re-adding the same email with add_contacts reactivates the contact but not the memberships. No email is sent.

delete_library_document

ChatGPT
Permanently deletes one Knowledge Library document, every version and its stored files. Without confirm:true the tool only returns a preview naming the document. Refused while any project links the document. Cannot be undone.

delete_persona

ChatGPT
Permanently delete a live persona from an active project. Trainee session history is preserved (snapshotted), but the persona itself cannot be recovered.

duplicate_project

ChatGPT
Creates a copy of a project (personas, goals, quiz, knowledge links) as a new project, optionally translated by AI into targetLanguage (language code such as fr or pt-BR). Without confirm:true the tool only returns a preview. The copy uses one project slot of the plan, runs in the background and cannot be edited until duplication completes; refused while the source still has documents processing or another copy is in progress.

generate_more_personas

ChatGPT
Generate 1-4 additional draft personas on top of the existing pool (max 20 total), optionally guided by instructions. Consumes one AI generation on access-code connections (3 max).

generate_more_quiz_questions

ChatGPT
Generate additional draft quiz questions on top of the existing pool (existing ones are never duplicated). Consumes one quiz generation on access-code connections (3 max).

generate_personas

ChatGPT
Start AI persona generation for a project from its knowledge base, the businessContext and the trainingGoals. Requires the knowledge to be ready. Goals steer the personas' scenarios and objections; without them generic template defaults apply. Asynchronous: the setup status reports personaGeneration=completed once the draft personas are available. The first full generation is free; on access-code connections each re-run consumes one of 3 AI generations. A re-run replaces the generated drafts (manually added drafts are kept).

generate_quiz

ChatGPT
AI-generate a draft quiz for the project from its knowledge and training goals; the questions stay drafts until saved. The first generation is free; on access-code connections each re-run consumes one of 3 quiz generations.

get_contact_import_job

ChatGPT
Status of a background contact import started by add_contacts with more than 50 rows: status, processed, created, updated, failed and invalid counts, timestamps and the error when it failed. Read-only.

get_crm_pipeline_digest

ChatGPT
Anonymized patterns from the closed deals of the CRM connection bound to a project: objection clusters, lost reasons, where deals stall, composite buyer profiles, won patterns. No names, no CRM records. connected=false when the project has no CRM; lowSignal=true when fewer than 15 lost deals had notes (shown for review, used for generation only when the manager opted in). Useful before suggest_training_goals or generate_personas to ground them in the organization's own pipeline.

get_live_context

ChatGPT
One Replio Live knowledge context in full: linked documents with their ready state, linked folders, the latest brief status (published, stale, objection approval, completeness, counts) and its cards: fact cards (id, kind, entity, attribute, value, unit, currency, period, conditions, hidden, manual, edited, quoteVerified), withheld cards with the reason, and objection cards (id, trigger, customerSays, approvedResponse, hidden). stillProcessing is true while a document is still being read or the brief is not built yet. With wait:true the call blocks up to about 25 seconds server-side and returns early when processing finishes.

get_live_workspace

ChatGPT
Replio Live overview for one organization: whether Live is enabled, the knowledge contexts (one per client, offer or language) with their linked document counts and brief publication state, the number of enrolled reps, and every Knowledge Library document and folder with the ids link_live_documents takes. Reps get access to every context automatically; a rep with several contexts picks one before the call. Answers available:false when Live is not enabled for the organization.

get_organization_usage

ChatGPT
Training minutes used, reserved by running sessions, remaining and the monthly pool (trial budgets are lifetime), seats assigned versus purchased, and the number of training projects (total and active).

get_phone_training

ChatGPT
Phone training (Replio Call) of one project: whether it is on, calls each participant may book per month, the end date, when the invitations went out, counts (participants with a number, consented, declined, calls booked this month) and every participant with name, email, number on file, consent, declined flag, bookings this month, next call and last outcome. Returns available:false with a reason when phone training is not activated on this Replio deployment.

get_project_kpis

ChatGPT
Key metrics for one training project: total and completed voice sessions, average AI score (0-10), trainee counts, and first-versus-recent progression across trainees.

get_project_report

ChatGPT
The data behind a project's printable report: headline stats (sessions, completed sessions, average score, trainees, progression), one row per trainee (status, sessions, average score, last session; first 50) and team skill averages with the trainees still under each target. CSV exports and the printable page are on the dashboard.

get_project_setup_status

ChatGPT
Setup pipeline state for one project: document ingestion statuses, whether the knowledge base is ready, persona generation progress, quiz state, current feature toggles, and the next step. With wait:true the call blocks up to about 25 seconds server-side and returns early when processing finishes; stillProcessing in the result says whether more processing is pending.

get_recent_activity

ChatGPT
Who trained and how it went over the last N days (default 1) for one project: session count, average score, per-trainee breakdown.

get_session

ChatGPT
One session of a project with its analysis: summary, strengths, improvements, score breakdown, the five skill scores, key moments, the trainee's goal, and the number of live coaching tips. Expert sessions return a recap (summary, key points, topics) instead of an analysis. No transcript.

get_session_transcript

ChatGPT
The conversation of one session as turns (trainee and persona) with timestamps, first 200 turns. Available on manager connections only; access codes cannot read transcripts.

get_trainee

ChatGPT
One trainee of a project: identity and status, aggregate stats (completed sessions, total duration, average AI score, sessions this week, progression, top strengths and improvement areas) and the 10 most recent sessions with score and outcome.

import_library_persona

ChatGPT
Copy a character from the organization's Persona Library into a project. A project in setup receives it in the draft persona pool (reviewable, survives regeneration, becomes live at activation); an active project gets a live persona immediately. Import is a copy: later library edits never change imported personas. The sheet is adapted to the project language automatically when the character's recorded language differs (same language stays verbatim, dialects included); translate overrides that: true forces adaptation, false forces verbatim. The imported persona speaks the project's language either way.

import_replio_document

ChatGPT
Copies one Replio-shipped document (docId from list_replio_library) into the organization's Knowledge Library under Replio's own folder and, when projectId is given, links it to that project with auto_sync. Extraction runs in the background and counts as one AI operation. A copy the organization already holds is returned as a duplicate.

invite_team_member

ChatGPT
Invites a manager to the organization by email: admin (full access to every project and setting) or viewer (read-only, limited to projectIds, at least one). Without confirm:true the tool only returns a preview. With confirm:true an invitation email goes out and the membership exists immediately. Viewer invitations are reserved to the organization owner; administrators can invite other administrators. Refused when the address is already a member.

invite_trainees

ChatGPT
Invite trainees to an active project by email: Replio sends a localized invitation email to each address. Refused while the project is still in setup. Invitees are saved in the organization's contact library, filed into a group named after the project. Max 100 per call; access codes may carry a total invitation cap set by the manager, and the error states the remaining budget when it is exceeded.

list_contact_groups

ChatGPT
List the organization's contact groups with id, name, description, color, member count and archive date. status: active (default), archived or all. Read-only.

list_contacts

ChatGPT
List the organization's contact library (people who can be invited to projects). Filters: search (email or name), status (active by default, deactivated, or all), groupId. Paged: page and limit (max 50). Each contact carries id, email, name, phone, status, the active groups it belongs to and how many projects it is enrolled in. Read-only.

list_generated_personas

ChatGPT
Full detail of every generated (draft) persona of a project in setup: personality, background, goals, difficulty, scenario and objections, exactly what the dashboard review screen shows.

list_knowledge_library

ChatGPT
Folders and documents of the organization's Knowledge Library: folders with id, name, path, parent and document count; documents with asset id, title, folder, extraction status and ready flag. Documents Replio ships for training are not listed. Optional folderId narrows the documents to one folder, search matches titles. Max 50 documents per call, most recently updated first.

list_library_personas

ChatGPT
The organization-level Persona Library: reusable characters (recurring customers, patients, candidates...) that can be copied into any training project. Each character records the language its sheet is written in. Requires write access.

list_personas

ChatGPT
Full detail of every live persona of an active project (name, role, personality, background, goals, difficulty, communication style, objections, including hidden coaching fields not shown to trainees). Draft personas of a project still in setup are not included. Requires write access.

list_project_templates

ChatGPT
Public training templates (sales, customer support, healthcare, soft skills, language learning...): id, name, slug, category and a short description. A template id is what create_project accepts as templateId.

list_project_trainees

ChatGPT
Trainees of one project with their invitation status (invited, active, deactivated), session counts, average AI score (0-10), last activity and how many personas they are restricted to. Paged, most recent connection first, max 50 per page; hasMore says whether another page exists. The membershipId returned here identifies the trainee in get_trainee, manage_trainee, set_trainee_personas and the phone training tools.

list_projects

ChatGPT
Training projects this connection can access, with organization, status and training domain. The project ids returned here are the input of every other tool.

list_quiz_questions

ChatGPT
Full detail of the project's draft quiz questions (question, options, correct answer, explanation, difficulty).

list_replio_library

ChatGPT
Documents Replio ships ready to use (for example irregular verbs for language learning): id, title, language, training domains, summary, and whether the organization already holds a copy and whether that copy is current.

list_sessions

ChatGPT
Voice sessions of one project over the last N days (default 7, max 90), newest first, one row per session: trainee, persona, mode, duration, AI score (1-10) with the five metric scores, evidence level, conversation outcome, scenario override, and whether the analysis is done. Expert sessions carry no score. Paged: hasMore and nextCursor continue the list.

list_team_members

ChatGPT
Managers of the organization (not trainees): email, name, role (owner, admin or viewer), membership date and, for viewers, the projects they can see.

manage_contact_group

ChatGPT
Create, update, archive or restore a contact group. create takes name (max 120 characters, unique among active groups) and an optional description; update takes groupId plus the fields to change; archive takes groupId and confirm, hides the group and keeps its members as contacts unless cleanupUnusedContacts:true, which also deactivates the members that belong to no other active group and are enrolled in no project (without confirm:true the tool only returns a preview); restore takes groupId, un-archives the group (renamed with a suffix if the name is taken again) and reactivates the contacts that the archive cleanup deactivated. No email is sent.

manage_knowledge_folder

ChatGPT
One folder operation in the Knowledge Library. create needs name (parentFolderId for a subfolder); rename needs folderId and name; move needs folderId and parentFolderId (omit it to move to the root); archive needs folderId and hides the folder from the library without touching its documents or their project links. Folder count is capped by the organization's plan.

manage_live_card

ChatGPT
Edit the fact cards of one Replio Live brief. update changes a card's hidden state, value, unit, currency, period or conditions (an empty string clears unit, currency or period); a hidden card never reaches reps, and both the hidden state and the edit survive rebuilds while the documents still give the same value. add writes a manual card (kind, entity, attribute, value, optional unit, currency, period, conditions, sourceLabel) that no document backs and that every rebuild keeps. delete removes a manual card only and requires confirm:true after a preview; a card read from a document cannot be deleted, only hidden. Card ids come from get_live_context.

manage_live_context

ChatGPT
Create, rename or archive a Replio Live knowledge context. create takes a name (1-80 characters) and an optional brief language (BCP 47, such as fr-FR); an organization holds at most 50 contexts and reps are granted the new context automatically. update changes the name and/or language; a language change rebuilds the brief in the background. archive hides the context from every rep (the documents stay in the library) and requires confirm:true after a preview; the last remaining context cannot be archived. Returns the context summary.

manage_live_objection

ChatGPT
Edit the objection cards of one Replio Live brief. update changes one card's hidden state, customerSays phrases (max 10) or approvedResponse (max 1000 characters); hidden state and rewording survive rebuilds. approve_all lets every visible objection card of the brief reach reps during calls; unapprove_all withdraws that. Objection cards are model wording and never auto-approve; every rebuild resets the approval. Ids come from get_live_context.

manage_project_documents

ChatGPT
Changes which Knowledge Library content feeds one training project and returns the project's links afterwards. link takes assetId (one document) or folderId (a folder, includeChildren default true for its subfolders) plus syncMode: auto_sync (default, follows new versions), manual_sync (moves only on a dashboard sync) or pinned (frozen on the current version). unlink takes linkId. The project's knowledge rebuilds in the background; refused while the project is being duplicated.

manage_trainee

ChatGPT
One action on one trainee of a project. resend_invitation emails a fresh login link (refused while deactivated). deactivate blocks new sessions, cancels the trainee's booked phone calls and keeps the account and history; the optional reason is shown to the trainee. reactivate restores access when a seat is free (refused at the seat limit) and sends no email. remove detaches the trainee from the project and frees the seat; account and history stay. send_reminder emails the practice reminder, at most once per 24 hours and never to a trainee who unsubscribed, is deactivated or whose project is not active. Every action except reactivate returns a preview without confirm:true and acts only with confirm:true.

regenerate_persona

ChatGPT
Replace one draft persona with a freshly generated one, optionally guided by instructions (for example a different role or industry). Consumes one AI generation on access-code connections (3 max).

regenerate_quiz_question

ChatGPT
Replace one draft quiz question with a freshly generated one. Consumes one quiz generation on access-code connections (3 max).

save_quiz

ChatGPT
Persist the draft quiz questions so trainees can take them (minimum 5, maximum 30). Omitting acceptedQuestionIds saves every draft question. Free: no AI generation consumed. Writes only to this project's data in Replio; no external system is contacted.

send_phone_invitations

ChatGPT
Emails the phone training invitation (booking link) to participants of a project whose phone training is on. scope all reaches every active participant who did not decline and marks the project so later newcomers are invited automatically; missing_number reaches only those without a phone number; memberships resends to the listed membershipIds. Without confirm:true the tool returns a preview with the number of recipients; with confirm:true the emails go out and the result gives sent and skipped counts. Returns available:false with a reason when phone training is not activated on this Replio deployment.

set_contact_group_members

ChatGPT
Add contacts to a group or remove them from it: groupId, action add or remove, contactIds (max 5000). Adding a contact that is already a member or removing one that is not is a no-op. Group membership does not enrol anyone in a project and sends no email.

set_live_brief_published

ChatGPT
Publish (published:true) or unpublish (published:false) one Replio Live brief. A published brief is what reps read during calls; hidden cards never reach them. Objection cards need a separate approval (manage_live_objection approve_all). briefId comes from get_live_context.

set_phone_training

ChatGPT
Turns phone training on or off for a project, sets how many calls each participant may book per month (1-20) and the last day of the programme (YYYY-MM-DD, null = open-ended; a past day is refused). Turning it off cancels every scheduled call; moving the end day earlier cancels the calls booked after it. Those two changes return a preview without confirm:true and act only with confirm:true; other changes apply directly. Turning it on sends no email. Returns available:false with a reason when phone training is not activated on this Replio deployment.

set_project_details

ChatGPT
Rename a training project and optionally set a short description.

set_project_features

ChatGPT
Enable or disable the project's optional training features. expertChatEnabled: 'Expert chat' voice mode where trainees ask free questions answered from the project knowledge (no roleplay, no scoring; consumes voice minutes like regular sessions). realtimeFeedbackEnabled: live coaching hints shown to the trainee during sessions. standaloneQuizzesEnabled: self-serve quiz practice area for trainees (only useful once a quiz is saved). hidePersonaDetails: trainees see only the persona's name, not role, difficulty or description (realistic cold-start scenarios). hideGoalSelection: skips the goal-selection step before sessions. Only the flags passed change; current values are part of the setup status.

set_trainee_personas

ChatGPT
Replaces the set of personas one trainee can practise with. personaIds must belong to the project (ids from list_personas); an empty array removes every restriction so the trainee sees all personas again. Takes effect immediately, no email is sent.

suggest_training_goals

ChatGPT
AI training-goal suggestions grounded in the project's linked Knowledge Library documents (same engine as the web wizard), available once the knowledge is ready. existingGoals excludes goals already agreed or already shown, so the suggestions are complementary. Read-only: nothing is stored.

update_contact

ChatGPT
Change a contact's name, phone or preferred locale in the organization's contact library. Only the fields passed change; an empty string clears a field. Email, status and project enrolments are not touched. No email is sent.

update_generated_persona

ChatGPT
Edit any field of a draft persona before activation (same as the dashboard review screen). Only the fields passed change. Free: no AI generation consumed.

update_persona

ChatGPT
Edit any field of a live persona on an active project; the persona keeps its sessions and history. Only the fields passed change. Refused on projects still in setup.

Capabilities

InteractiveWrites

Example Prompts

Click any prompt to copy it.

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/plugin_asdk_app_6a53d8f876808191ab508c8262a1d318"><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/plugin_asdk_app_6a53d8f876808191ab508c8262a1d318)