MCP App Store
by KeenEthics
Scalign icon

Scalign MCP Server for Claude and ChatGPT

By Scalign AS

Overview

Scalign is sales call intelligence for B2B sales teams: it helps you prepare for customer calls, understand your deals and follow up. This connector lets you ask Claude about your accounts, deals and contacts, using the sales calls your team has recorded with Scalign and the insights Scalign has extracted from them. Ask in plain language: - Brief me on a company before the next call: what was said, what is open, what changed since last time. - Who is on the buying committee, what role each person plays, and which roles are not yet covered. - How healthy is this deal, and what is the customer's reason to buy. - Which objections come up most often across deals, with the customer's own words. - What product feedback are customers giving us. - My open follow-up actions, and mark them done. - What a specific person said, with the transcript. What the connector does - Reads companies, contacts, buying committees, deal health, signals and reasons to buy, and the deal mapped onto your sales methodology (e.g. MEDDPICC). - Reads call recordings, summaries and transcripts, and customer email threads when email sync is on. - Reads objection and product-feedback clusters and the quotes behind them. - Lists follow-up actions and marks them done. - Imports a meeting transcript from another tool so Scalign analyses it. - Reads your ideal customer profile; workspace admins can add attributes and change scores. - Answers how-to questions about Scalign from its help articles. - Sends feedback to the Scalign team. It cannot record, delete or edit meetings, or send email or messages. Access is the Scalign workspace and permissions of the account you connect. You need a Scalign workspace with at least one recorded call. Sign in with the account you use at app.scalign.ai.

How to connect

  1. Claude: Open the Scalign listing in the Claude connectors directory and choose Connect. Claude listing Scalign exposes 30 tools here.
  2. ChatGPT: Open the Scalign listing in the ChatGPT app directory and choose Connect. ChatGPT listing Scalign exposes 35 tools here.

Tools

add_icp_attribute

ChatGPT
Create a new ICP scoring attribute (e.g. 'Company Size', 'Tech Stack') with its possible values. Optionally assign each value a fit category (ideal, not_ideal, no_go); values left without a category are neutral. Admin-only. To score values on an existing attribute, use set_icp_attribute_score.

agent_feedback

ChatGPT
Report friction you (the AI agent) hit using Scalign's tools — safe and side-effect free, it only sends a note to the Scalign team and needs nothing from the user: a tool whose description misled you about what it does or returns, a result that was wrong, empty or hard to use, a call you had to retry or work around, a tool that was missing for what the user asked, or an idea that would make these tools work better for an agent. Fire it as soon as the friction happens, before moving on — this is the Scalign team's only view of how tools behave in practice. Describe what you called, what you expected and what you got; name the tool. Keep the user's data out: no transcript excerpts, emails, names or company details. Not for the user's own feedback about Scalign — that is submit_feedback.

create_note

ChatGPT
Save a note from the user about a customer — what happened in a meeting, call or conversation Scalign didn't record. Scalign works out which company it's about (or asks the user in the app when unsure), captures any commitments as actions, and updates the deal. The company is attached from its name in the text.

finalize_call_recording

ChatGPT
Save the user's answers for a recording: the meeting type, which voice is whom, and people who were in the room but not on the invite. A held recording (needs_finalizing) is all-or-nothing: one call with the meeting type and every non-brief voice — proposals included, by repeating the proposed email — and, for a sales type, a buyer with a company email; anything less saves nothing and the error says what is missing. Once released, saves can be partial. effect says what the save set off: released or reanalyse — the summary, actions and stakeholders are recalculated and replace the current ones within a few minutes; reattribute — quotes re-labelled, nothing else changes; pending — folded into a run already queued; none — nothing changed. Language, diarization and transcript switching stay in the app.

get_buying_committee

ChatGPT
The company-level map of who is in the deal — the buying committee for one company (by domain). B2B deals are sold to a group, and any one member can stop the deal: this names them. Returns the highest-leverage move (the single cross-committee thing to do next) and its reasoning; the members (every external human who's been on a call) each with their buying role, health, one-line summary, reasoning, recommended next action, and evidence quotes; role gaps (required roles no one covers — an open Economic Buyer or Ratifier gap usually means a deal at risk); and named absentees (people mentioned but never met, often with no email yet). For one person's deep read use get_stakeholder; for the buyer arguments (the 'why') use get_company_three_ys.

get_call_recording

ChatGPT
Get full details for a specific call recording including summary, classification, participants, action items, objections, and product feedback. source is the capture path — recall (meeting assistant), recall_desktop (desktop app), import (brought in with import_call_recording), or a third-party import (fathom, fireflies, fellow, attio, modjo) — and is null when no transcript has landed yet. Transcript quality and participant metadata differ by source, so check it first when triaging a quality complaint.

get_call_recording_transcript

ChatGPT
Get the full transcript of a call recording as a list of speaker turns with timestamps. Each turn's speaker is who the recording resolved that voice to; transcript_label is the capture provider's raw label, kept for reference. The email is set only when the voice is linked to a participant: with a null email the voice is not placed on anyone, and any name is just a guess heard in the meeting or the raw label.

get_company

ChatGPT
The default for briefing intent — 'brief me on this account', 'is this a good fit', 'why is this deal stalling'. Returns an overview of a single company (account): basic facts, a meeting-activity summary (count, first and last meeting dates), and its Reasons to Buy. Reasons to Buy are the synthesized buyer arguments for a deal — each with situation, pain, impact, solution, desired outcome, objections, next step, and state (open/won/lost/abandoned). For closing or positioning questions ('how do I close them', 'what's the pitch', 'why should they buy'), start with get_company_three_ys instead. For the full list of the company's meetings, use list_call_recordings with the domain filter. For ICP fit, use get_company_icp_fit. For buying signals, compelling events and blockers (deal momentum), use get_company_signals.

get_company_deal_health

ChatGPT
Use when the question is how a deal is doing, how likely it is to close, or what is weakest about it (e.g. 'how healthy is this deal', 'what's the score', 'what should I shore up'). Returns the company's most recent deal-health snapshot: composite score and band, the headline, what's strongest, weakest and missing, and each pillar with its score, note and the meeting quotes behind it. previous_composite is the score before this snapshot and change_note says what moved it. For what to do about it, use list_actions — moves live in the action layer, not here. Read-only — it never recomputes, so a company that has no snapshot yet returns one the rep can trigger by opening the company in the app.

get_company_icp_fit

ChatGPT
Returns how a specific company scores against the workspace's ICP criteria (see list_icp_attributes for the attributes and their scoring). Each attribute includes its detected values grouped by fit category (ideal, not_ideal, no_go, neutral), a confidence level (low, medium, high), and reasoning explaining how the values were determined.

get_company_signals

ChatGPT
Account-level signals that this company is moving: buying signals and compelling events, aggregated across all of the company's meetings. Buying signals are contextual indicators that raise buying likelihood (e.g. a new VP hire, funding round, market expansion). Compelling events are the Why-Now triggers — time-bound or change-driven situations creating pressure to act this quarter — deduplicated across meetings and ordered active → upcoming → resolved, each with its source meetings and quotes. Blockers are the gates in the buyer's process holding the deal without the buyer doubting us (an approval, an internal decision, budget timing, a pause, an outside dependency), open first, with who holds each and when it was last mentioned; 'unconfirmed' means past its lift date or unmentioned for two months, so check it still holds. Use this to gauge deal momentum before a call and to answer what is in the way. For the speakable Why Change / Why Now / Why Us framing built partly on these, use get_company_three_ys.

get_company_three_ys

ChatGPT
Start here when the question is about how to close, pitch, or position a deal (e.g. 'how do I close them', 'what's the pitch', 'why should they buy'). Synthesizes the '3 Ys' for a company: Why Change (the buyer's cost of staying put), Why Now (the closing window), and Why <seller> (why the seller is the right move). Each is a short list of speakable bullet points a rep can walk into a meeting with, derived live from the company's open Reasons to Buy, compelling events, and named alternatives. Returns empty lists when there isn't enough signal yet — for low-signal accounts, fall back to get_company. Synthesized live on demand, so allow a moment for the first call; results are cached afterward.

get_help_article

ChatGPT
Read one Scalign help article in full, as markdown: steps, permissions, troubleshooting. Get the slug from list_help_articles first — never guess one. Answer from the article, not from memory: keep its steps and menu paths as written (e.g. Settings → Connections), include only the part that answers the question, and link the article's url so the user can read the rest. If no article covers the question, say so and point to support@scalign.ai instead of improvising how Scalign works.

get_meeting_details

ChatGPT
What Finalize asks about one recording: the meeting type, the roster, and every voice on the transcript with its talk share, longest quotes, and either answer (confirmed), proposed (the system's guess with its evidence, for the user to confirm) or neither (open). colleagues lists workspace members not on the roster, for a voice that turns out to be one of them. Use it to ask the user who a voice is before finalize_call_recording (personal tokens; a workspace token sends them to the link instead); when the quotes aren't enough, get_call_recording_transcript has every turn. finalize_url opens the same item in the app, with audio and video.

get_objection_cluster

ChatGPT
Get detailed objections for a specific objection cluster. Returns each individual objection with its reasoning, topics, source call recording, and the external company domain that raised it. Items are ordered by relevance to the cluster theme.

get_product_feedback_cluster

ChatGPT
Get detailed feedback items for a specific product feedback cluster. Returns each individual feedback with its sentiment, reasoning, source call recording, and the external company domain that raised it. Items are ordered by relevance to the cluster theme.

get_product_update

ChatGPT
Read one Scalign product update in full, as markdown with its url: each change and why it matters, as officially announced. Dates come from list_product_updates. Answer from the update, never from memory; link its url.

get_stakeholder

ChatGPT
The full read on one buyer-side person, keyed by email. Returns durable identity (name, title, professional/core interests), their deal read (buying role, health/stance, a one-line summary, the reasoning behind it, and the recommended next action on them), the dated health trend, and the verbatim evidence quotes grounding the read. Role and health are null when there's no read yet (a silent attendee). For the whole company's committee, use get_buying_committee; to find a person's email, use list_stakeholders.

import_call_recording

ChatGPT
Import a meeting transcript Scalign did not record — one the user pastes, exports from another tool, or already has as text — so it is analysed like any recorded meeting: summary, actions, objections, stakeholders, Reasons to Buy, deal health. Every turn needs the speaker's name, email, and start and end in seconds from the start of the meeting; the email is what places the voice on a person, so when the source doesn't give one, ask the user who each speaker is before importing rather than guessing or leaving it out. Set meeting_url to the real meeting link when there is one, so a transcript of a meeting Scalign also recorded merges with it instead of duplicating. You own the recording. Importing the same transcript twice returns the existing recording. Processing is asynchronous: the result lands within a few minutes; read it with get_call_recording.

list_actions

ChatGPT
List action items (to-dos) captured from meetings, customer email and the user's notes, generated per deal, or added manually. Defaults to open actions, newest first. Every row carries a source (call | email | note | null for manual/company-scope) with the meeting or email thread it came from. Scope with mine (your own to-do list), domain (one company's actions), source, or call_recording_id (one call's actions), and narrow by status or creation date range. To read the thread behind an email-sourced row, call get_company_emails on its company_domain and match email_subject.

list_call_recordings

ChatGPT
Search and list call recordings with their title, date, participants, classification, summary, and source (capture path: recall meeting assistant, recall_desktop desktop app, import brought in over MCP, or a third-party import; null until a transcript lands).

list_companies

ChatGPT
The entrypoint for working with accounts. Lists the workspace's companies with basic facts (name, domain, meeting count, last meeting date, last_activity_at), newest activity first. Optionally narrow by free-text search (name or domain) and/or by ICP attribute filters. last_activity_at is when the deal last moved — a meeting, a synced email, a note, or a re-read of Reasons to Buy, buying committee, deal health, blockers or compelling events; enrichment and renames don't count. For a recurring job that needs only what changed since its last run (e.g. a scheduled sync of the methodology read into a CRM), use list_company_updates instead. From a row, drill in with get_company (overview + Reasons to Buy), get_company_icp_fit (ICP scoring), get_company_signals (buying signals, compelling events + blockers), get_company_three_ys (Why Change / Why Now / Why Us), get_company_methodology (the deal in the workspace's sales methodology), or list_call_recordings (the company's meetings, via the domain filter).

list_company_updates

ChatGPT
The change feed for recurring jobs (e.g. a scheduled sync of the methodology read into a CRM): the companies whose deal data changed, in the order the change reached Scalign, from a cursor on. A change is a meeting, a synced email, a note, or a re-read of Reasons to Buy, buying committee, deal health, blockers or compelling events; enrichment and renames don't count, and neither do changes made in the CRM itself (stage, amount) — run without a cursor now and then (e.g. weekly) to refresh everything. Rows carry last_changed_at (when the change reached Scalign — an old meeting imported today is a change today) next to last_activity_at (when it happened). Each page returns companies (as list_companies, plus last_changed_at), next_cursor (where this page ended) and has_more; while has_more is true, call again with next_cursor. A job that can keep state persists the last page's next_cursor as the next run's start; one that cannot (a scheduled task with no memory) passes since instead — the last run's time with some overlap, e.g. 25 hours ago for a daily task. Pass neither for the first run or a full refresh. Then read each company with get_company_methodology, get_company or get_company_signals.

list_help_articles

ChatGPT
List Scalign's help articles: the official guides to setting up and using Scalign itself. Use it whenever the user asks how Scalign works, how to set something up, what Scalign can access, or why something in Scalign isn't working — e.g. connecting a calendar so the meeting assistant joins (Google, Outlook/Teams), Microsoft admin consent for IT, connecting Gmail/Outlook for follow-up drafts, connecting a CRM (HubSpot, Salesforce, Pipedrive, Attio) and what Scalign writes to it, Slack notifications, connecting an AI tool over MCP or creating access tokens, importing meetings from another note-taker, the permissions Scalign asks for, or how a rep uses Scalign day to day. Do NOT use it for the user's own data (meetings, companies, deals, people, insights) — the other Scalign tools answer those — nor for general sales advice, nor to report a bug (submit_feedback). Returns slug, title, summary, tags and url per article. Pick the article(s) that fit by title and summary, then read them with get_help_article; the summary alone is not enough to answer from.

list_icp_attributes

ChatGPT
Returns the workspace's Ideal Customer Profile (ICP) scoring criteria — the definition, not any company. Each attribute (e.g. Company Size, Sales Team Size, Business Type) lists its possible values grouped by fit category: ideal, not_ideal, no_go, or neutral. Neutral values have no score assigned. Empty categories are omitted. To see how a specific company scores against it, use get_company_icp_fit; to filter companies by it, use list_companies.

list_objection_clusters

ChatGPT
List clustered sales objections extracted from call recordings. Each cluster groups related objections and includes the total count and when it was last mentioned.

list_product_feedback_clusters

ChatGPT
List clustered product feedback themes extracted from call recordings. Each cluster groups related feedback items and includes sentiment breakdown (positive, negative, suggestion) and when it was last mentioned.

list_product_updates

ChatGPT
List Scalign's product updates, newest first: what shipped and when, as officially announced. Use it whenever the user asks what's new in Scalign, whether or when a feature shipped or changed (e.g. 'can Scalign record video yet?', 'when did notes start going to the deal?'), or what changed recently that could explain something behaving differently. Not for how to set up or use a feature (list_help_articles), nor for the workspace's own data. Returns date, title, highlights (one line per change in the update) and url per update; the full text comes from get_product_update when the highlights don't answer the question.

list_quotes

ChatGPT
Search and list quotes from call recordings. Quotes are verbatim excerpts from external meeting participants (prospects, customers), each attached to an insight (painpoint, objection, product feedback, buying signal, customer value, or alternative). Useful for finding authentic customer language, testimonials, and voice-of-customer content.

list_stakeholders

ChatGPT
Lists buyer-side people (stakeholders) the workspace has met, newest contact first. A stakeholder is one external human in the context of one deal. Optionally narrow by free-text search (name, email, or company), by a single company (domain), by buying role (champion, economic_buyer, decision_maker, technical_buyer, ratifier, …), or by health (positive / neutral / sceptic). Use 'unknown' for role or health to find people with no read yet. Each row carries name, title, buying role, health, company, and last contact date. For one company's full committee (members + role gaps + missing absentees + the highest-leverage move), use get_buying_committee. For a single person's full read with evidence quotes, use get_stakeholder.

remove_assistant_from_meeting

ChatGPT
Stop the Scalign meeting assistant from recording one of the user's calendar meetings. For an upcoming meeting it cancels the assistant (and keeps auto-join from booking it again); if the assistant is already in the call, it leaves and recording stops for the rest of the meeting. status says which: cancelled, left_call, or already_left.

send_assistant_to_meeting

ChatGPT
Send the Scalign meeting assistant to one of the user's calendar meetings to record and transcribe it. For an upcoming meeting it joins shortly before the start; for one in progress it joins right away. Overrides the user's auto-join settings for this meeting.

set_action_status

ChatGPT
Complete, dismiss, or reopen an action item: completed = done, dismissed = deliberately not doing it, open = undo any terminal state. Completing and dismissing are attributed to you; reopening clears the attribution.

set_icp_attribute_score

ChatGPT
Change the fit category of one value on an existing ICP attribute — e.g. move 'Company Size: 1-50' from ideal to not_ideal. Pass category 'neutral' to clear the score. Attribute and value are addressed by their names/labels (see list_icp_attributes). Admin-only.

submit_feedback

ChatGPT
Send the user's product feedback about Scalign itself — a bug, an annoyance, a feature request, or praise — to the Scalign team. Use it when the user says something is broken, missing, confusing, or wishes Scalign did something. Do not use it for feedback their customers gave about the user's own product (that's product feedback insights from calls). Confirm with the user before sending, and pass their words, not a paraphrase.

add_icp_attribute

Claude

create_note

Claude

get_buying_committee

Claude

get_call_recording

Claude

get_call_recording_transcript

Claude

get_company

Claude

get_company_deal_health

Claude

get_company_emails

Claude

get_company_icp_fit

Claude

get_company_methodology

Claude

get_company_signals

Claude

get_company_three_ys

Claude

get_help_article

Claude

get_objection_cluster

Claude

get_product_feedback_cluster

Claude

get_stakeholder

Claude

import_call_recording

Claude

list_actions

Claude

list_call_recordings

Claude

list_companies

Claude

list_company_updates

Claude

list_help_articles

Claude

list_icp_attributes

Claude

list_objection_clusters

Claude

list_product_feedback_clusters

Claude

list_quotes

Claude

list_stakeholders

Claude

set_action_status

Claude

set_icp_attribute_score

Claude

submit_feedback

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/scalign"><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/scalign)