MCP App Store
by KeenEthics
Convin Sense icon

Convin Sense MCP Server for ChatGPT

By Convin

Overview

Bring your Convin Sense data into ChatGPT. Ask about your campaigns, leads and AI agents in plain language: how campaigns performed over a period, why leads are or aren't interested, what was said on a call, how your agents are set up, and how much calling capacity and usage your organization has. Access is read-only and limited to your own organization. Your role's permissions and your organization's PII masking settings apply to every answer. Requires an existing Convin Sense account with ChatGPT access enabled by your administrator.

How to connect

  1. ChatGPT: Open the Convin Sense listing in the ChatGPT app directory and choose Connect. ChatGPT listing Convin Sense exposes 29 tools here.

Tools

get_agent

ChatGPT
One AI agent: business outcome, languages, channels, gender, voice, opening message per call channel, attached tools and the campaigns that use it. The published prompt's length and whether unpublished edits exist are always reported; include prompt adds its text, which is long. Never includes credentials.

get_billing

ChatGPT
The organization's billing account: plan and subscription status, billing cycle and current period dates, concurrent call and contact limits, wallet balance with an estimated runway, the current period's charges so far per billable metric, and the last few invoices. With past_periods it also returns that many closed billing periods before the current one (up to a year), each with what it was invoiced per billable metric and its invoice's payment status. Blocks marked invoiced come from the billing provider. Call it for "what plan am I on", "what is my balance", "what have I been charged this cycle" or "what was I billed last month"; for consumption and charges of one campaign over a chosen period, use get_usage.

get_call_outcomes

ChatGPT
Why calls do or don't connect in a period, for the campaign_ids given or, with none, every campaign in the organization: attempted, connected and failed calls with the failure split (DND, busy, no answer, cancelled, error), the same per caller id (the numbers calls are placed from, so a blocked or spam-flagged number stands out), and a compact best-time-to-call summary per calling channel: connect rate by time of day plus the best and worst day-and-slot cells, read only with include best_times (the heaviest read; ask for it when the question is about when to call). Use it after get_campaign_funnel shows a low connect rate. The period is on the call time unless period_by is lead_created; with no period it reports yesterday. A period spans at most 90 days (calendar days in Asia/Kolkata).

get_campaign

ChatGPT
One campaign in full: its goal (the agent's business outcome, which Goal Achievement is judged against), type, direction, channels, agent, language, schedule and calling windows, lifecycle status, lead counts by state and by interest level, and plain-language health signals. Contains no prompts, messages or credentials. For what the campaign measures and how each metric is defined (metrics, entities, dispositions) use get_campaign_dimensions.

get_campaign_dimensions

ChatGPT
What one campaign measures and how: its AI metrics with their allowed values and the definition each is evaluated against (the campaign's goal for Goal Achievement, the scoring criteria for Interest Level, Lead Qualification and custom metrics), the entities extracted from conversations, and its call disposition templates with their L1/L2/L3 values. Call it before get_lead_breakdown, list_evaluation_reasons or list_lead_summaries so the metric names and values you pass are the campaign's own, and to explain what hot, qualified or goal achieved mean for this campaign.

get_campaign_funnel

ChatGPT
The analytics dashboard for a period, pooled over the campaign_ids given or, with none, every campaign in the organization: leads attempted and connected with the connect rate, call attempts, connected calls and talk time, message attempts, touchpoints, and the interested and goal-achieved counts, each with its change against the previous period of the same length. bucket (day, week or month) and group_by campaign add rows that split the call and message counts per period and per campaign, so a week-on-week trend or a comparison of campaigns is one call: never call this tool once per campaign or once per week. By default it reads the calls, messages and leads attempted and connected; include trends adds the change against the previous period and include outcomes the interested and goal-achieved counts, each at the cost of more reads, so ask for them only when the question needs them. A read that does not finish in time is reported in notes while the rest of the answer stands. Use get_call_outcomes for why calls did not connect and get_lead_breakdown for counts per state or interest. With no period it reports yesterday, as the dashboard does by default; period_by lead_created follows the leads created in the period instead of the activity in it. A period spans at most 90 days (calendar days in Asia/Kolkata; weeks start on Monday).

get_capacity

ChatGPT
Live calling capacity right now: how many concurrent AI-call slots the plan allows, how many are in use (connected, still dialing, or stuck after the call ended), how many are free, how many calls are waiting for a slot, which campaigns hold slots, and the widget call pool's cap and usage, with plain-language signals. Use it when calls seem slow to start or queue up, or when asked whether the organization is at capacity. Given a campaign it adds that campaign's own slot usage: slots held and since when, calls waiting and any reserved allocation. Figures are a snapshot at as_of; a section with available=false could not be read and its numbers are unknown, not zero.

get_context

ChatGPT
Who is connected: the user and their role, the organization, which PII masking applies to the other tools' answers, and the scopes this connection was granted. Call it first in a conversation.

get_interaction

ChatGPT
One interaction in full: channel, direction, status, timestamps, call duration and attempt, the message text or call transcript (PII masking follows the organization's settings), the provider's call summary, AI dispositions with reasons, and the delivery error of a failed message.

get_lead

ChatGPT
One lead's profile: contact details (masked per the organization's PII settings), campaign, state, owner, interaction counts and the AI evaluations with reasons. include adds the conversation summary, extracted entities, CRM fields and pending actions. For the conversation history use get_lead_timeline.

get_lead_breakdown

ChatGPT
Counts leads per value of one dimension (state, interest, qualification, goal, a custom metric, or AI call disposition) for the campaign_ids given or, with none, every campaign in the organization: the fastest answer to questions about many leads, in one call rather than one per campaign. Metric dimensions count connected leads (answered a call or replied), as on the campaign analytics dashboard, over an optional period on lead creation or activity; disposition counts classified calls over a required call-time period and needs exactly one campaign. Any period spans at most 90 days (calendar days in Asia/Kolkata). Follow up with list_evaluation_reasons for the reasons behind a value.

get_lead_timeline

ChatGPT
A lead's conversation history, newest first: AI and human calls, WhatsApp, SMS and email, each with direction, status, call duration and AI disposition. Answers "what happened with this lead". Open one entry with get_interaction to read the message or transcript.

get_profile

ChatGPT
The connected account in brief: a stable account id, the user's name and email, and a nickname naming the organization. For role, PII masking or granted scopes use get_context instead.

get_role

ChatGPT
One role in full: description, whether it is a built-in system role, how many users hold it, and its permissions grouped by app area with each permission's label and description, next to the catalog permissions it does not grant. Use it to answer what a role can and cannot do; through any role it also shows the complete permissions catalog.

get_usage

ChatGPT
What one campaign consumed over a period and what it cost, grouped one way at a time. Rows are computed from recorded usage at the plan's current unit prices (the Campaign Costs page's own math), not read from invoices; grouping by metric also attaches the billing provider's invoiced charges for the current billing period, organization-wide, for comparison. Lead charges count leads charged in the period (each lead once, on its first contact attempt or at upload), not leads uploaded, and are gross: the plan's included leads apply to the whole organization per billing cycle. Use it to explain a campaign's bill from activity; for the account state use get_billing. A period spans at most 90 days (calendar days in Asia/Kolkata).

list_agents

ChatGPT
The organization's AI agents with their languages, channels, voice and whether they are active, most recently updated first. The agent a campaign uses is named in get_campaign. For prompts, opening messages, tools and campaigns use get_agent.

list_campaigns

ChatGPT
The organization's campaigns, most recently updated first, each with lead counts by interest level. Use it to find a campaign id or to compare campaigns at a glance (which have the most hot leads). For one campaign's setup, status and health use get_campaign.

list_evaluation_reasons

ChatGPT
The reasons the AI recorded for the leads in one bucket of a metric (for example Interest Level = not_interested), newest first with lead ids. Read these to explain why leads landed in a bucket instead of opening leads one by one.

list_integrations

ChatGPT
Every provider the organization has configured, in one call: messaging (WhatsApp, SMS, email), telephony, the voicebot calling infrastructure, AI models (LLM, TTS, STT) and lead sync, each with its provider, channel, enabled state and whether it runs on Convin's managed account. Contains no credentials, endpoints or settings. Use it to answer which channels or providers are set up and to find an integration id; for the numbers behind telephony and WhatsApp integrations use list_phone_numbers, for WhatsApp templates use list_whatsapp_templates.

list_lead_stages

ChatGPT
The organization's lead stages in funnel order, each with its colour, whether it is an active or terminal stage and whether it is still in use. Read it before filtering or reporting on leads by stage so the stage names you use are the organization's own.

list_lead_summaries

ChatGPT
AI conversation summaries for many leads of one campaign in one call, most recently active first, each with the lead's state, interest level and last activity. Use it to read what leads are saying at scale. total is capped at 1000. A from/to window on last activity spans at most 90 days.

list_lead_uploads

ChatGPT
The bulk lead uploads made to one campaign, newest first: status, file name, who uploaded it, and the total, processed, successful and failed row counts with the upload, start and completion times. Use it to see how many leads were added when and whether an upload is still running or failed; leads added one at a time or through an integration are not listed. Files and download links are never returned. A from/to window on upload time spans at most 90 days.

list_phone_numbers

ChatGPT
Every number the organization calls or messages from, in one list: telephony caller ids per telephony integration and WhatsApp business numbers per WhatsApp integration, each with type, status, the integration it belongs to and the campaigns using it (for WhatsApp numbers, the inbound campaigns answering on it). Use it to find which number a campaign dials from, whether a number is in use, or which numbers a provider has. For the integrations themselves use list_integrations.

list_roles

ChatGPT
Every role in the organization, built-in and custom, with its description, how many users hold it, how many permissions it grants and the app modules it hides. Include permissions adds each role's permissions as resource:action pairs, the whole permissions matrix in one call. For one role's permissions with labels, and the catalog permissions it lacks, use get_role.

list_users

ChatGPT
The organization's users with name, email, role and status, plus pending invitations as rows with status invited, newest first. Use it to find who has access, who holds a role, when someone last signed in, or whether a person has been invited yet. For what a role allows use get_role.

list_variables

ChatGPT
The organization's variable definitions: the keys that prompts, message templates and lead uploads refer to, each with its name, description, data type and whether every lead must have a value. To find one variable, search by its key. Definitions only: a variable's values are never returned.

list_voices

ChatGPT
The organization's cloned voices, the custom voices its AI agents can speak with: name, language, gender, provider and whether the voice catalog also shares it with other organizations. Metadata only, never audio. The voice an agent uses is named in list_agents and get_agent.

list_whatsapp_templates

ChatGPT
The organization's WhatsApp message templates as last recorded from the provider: name, language, category, Meta approval status, quality score and reject reason. Use it to check which templates are approved and sendable, or why one is not. It reads stored statuses and never contacts the provider, and it does not say which campaigns use a template.

search_leads

ChatGPT
Find leads by an exact identifier (full email address, full phone number or external id), or list them by campaign, state and interest, newest first. Leads cannot be searched by name or by partial values. For counts use get_campaign; for what many leads are saying or why they are not interested use get_lead_breakdown and list_lead_summaries rather than paging. Contact details follow the organization's PII masking.

Capabilities

Campaign PerformanceLead Search And HistoryCall Transcripts And SummariesAi Agent SetupUsage And Capacity

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/convin-sense"><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/convin-sense)