MCP App Store
by KeenEthics

Overview

Give Claude access to customer conversations and next-generation revenue intelligence behind your deals. Aircover turns meetings, transcripts, CRM context, qualification, account plans, research, and AI-generated insights into a rich understanding of every customer relationship. Aircover’s Account & Deal Graph connects conversations, stakeholders, priorities, products, objections, commitments, and deal activity over time, so Claude understands the history behind each opportunity, not just individual records. Ask a question and get an answer grounded in what your buyers actually said, not just what was entered into the CRM. Ask Claude: • “What changed in this deal since our last call?” • “Which deals are at risk, and why?” • “Prep me for my calls this week.” • “Update this account plan based on the last three meetings.” With Aircover, Claude can: • Understand customer conversations: Search meetings and transcripts for priorities, objections, commitments, competitive signals, and next steps. • Connect conversations to deals: Use the Account & Deal Graph to connect what buyers said with qualification, stakeholders, products, and account history. • Reason across calls: Run Aircover agents and research to find patterns, risks, coaching opportunities, and Voice of Customer insights. • Work with account context: Read and update Account Plans, Mutual Action Plans, and custom deal documents with version history. • Tap into sales knowledge: Query Aircover research and connected content from Google Drive, OneDrive, SharePoint, and uploaded documents with cited answers. Aircover complements your CRM by giving Claude the conversation intelligence and relationship graph behind the fields: the customer’s words, the evidence behind qualification, who’s involved, what changed, and how the relationship evolved. Connect with OAuth. Existing roles and permissions are respected, and admins can control MCP write access and client approvals. Requires an Aircover account.

Tools

agent_results

ChatGPT
Run an agent against a meeting's transcript and return its structured result. Mirrors the meeting-details page: a cache hit short-circuits (fast, cheap), otherwise the agent runs against the transcript and the result is saved on the meeting, replacing any earlier result for that agent. Set refresh to force a re-run.

ask_documents

ChatGPT
Ask a natural-language question about your organization's uploaded sales documentation (Google Drive / OneDrive / SharePoint / manually-uploaded files). Returns an LLM-generated answer with source citations. Does NOT search meeting transcripts, deal records, or CRM data — use get_meeting_transcript / get_deal for those.

get_account_deal_agent_results

ChatGPT
Read an account-level or deal-level agent document for one deal + template. These are Aircover's account and deal agents: an Account Plan or a Mutual Action Plan (both ACCOUNT level — the document is about the account), or a custom Deal Agent (DEAL level — one opportunity). They run against a WHOLE account or deal, not a single meeting, and their output is a versioned document — this returns the current version's contents by default, or a specific historical version with version. A deal that has never had this agent run comes back with an empty status, current_version 0, an empty versions array and an empty result: that means 'no document yet', not an error. declared_sections lists the section ids, titles and author guidance this document is made of. If save_account_deal_agent_document is in your tool list, pass those ids to it to write a new version, and pass this call's current_version as its base_version; if it is NOT listed, an Aircover admin has turned MCP writes off for this organization, and can turn them back on with the 'Allow MCP Clients to Write' switch on the MCP settings page. Running an agent, restoring a version and CRM-syncing still happen in the Aircover web app. Call the deal by the deal_name in the response; deal_key is an internal storage key, returned only so you can pass it to another tool, and must never appear in what you write for a person.

get_deal

ChatGPT
Fetch a deal by prospect_org + deal_id within the caller's customer org.

get_deals

ChatGPT
List deals in the caller's org — the same data the deal dashboard table shows. Returns ONLY cached data; if the cache is missing, returns an error instructing the caller to retry. Sort, filter, and pagination are applied over the full filtered dataset, not just one page, so sort=sfdcOppAmount desc limit=25 reliably returns the top-25-by-amount across the whole org. Viewers can only query their own deals (pass rep or leave it blank). Use get_teams to translate team names to IDs before passing to team_ids.

get_meeting

ChatGPT
Fetch a single meeting's details by ID, including attendee split, notes, and customer summary. You can look up the meeting either by meeting_id OR by the (start_time, room) pair — provide one or the other.

get_meeting_transcript

ChatGPT
Fetch a formatted transcript for a meeting, optionally clipped to a time window.

get_qualification_results

ChatGPT
Fetch the per-dimension qualification breakdown (e.g. MEDDPICC dimensions, coaching notes, best-meeting scores) for a single deal + template. Mirrors the data shown in the deal-details drill-down. Any authenticated user in the org may read; get_deals is where discovery is constrained by role. Call the deal by the deal_name in the response; deal_id is an internal storage key, returned only so you can pass it to another tool, and must never appear in what you write for a person.

get_teams

ChatGPT
List active (non-deleted) teams in the caller's org as {team_id, name} pairs. Use this to translate human-readable team names into the team_ids accepted by get_deals.

list_account_deal_agents

ChatGPT
List the account-level and deal-level agent templates available to the caller, grouped into the three categories: account_plan and mutual_action_plan (both ACCOUNT level) and deal_agent (DEAL level), each with its default template. Both scopes are covered by this one tool. Use it to find a template_id for get_account_deal_agent_results. All three categories are always returned; a category the org has no templates for comes back with an empty templates array and an empty default_template_id. For per-meeting agents use list_agents instead.

list_agents

ChatGPT
List agents (coaching-style prompt templates) available to the caller, optionally filtered by category.

list_meetings

ChatGPT
List meetings in a date range for the caller or another user in the org, optionally narrowed the way the /meetings page narrows them. Returns lean per-meeting summaries.

list_reports

ChatGPT
List reports available to the caller. Returns report ID, title, status, and last-updated timestamp. When the user has organized reports into folders, results are grouped by folder; otherwise a flat list is returned sorted by most recently updated. Use the returned report IDs with read_report to fetch full report data.

read_report

ChatGPT
Read the data for a previously generated report. Returns either the generated report results (LLM-produced summary, sections, and insights) or the raw agent data (per-meeting coaching scores). When data_type is omitted, returns results if available, otherwise falls back to raw data.

save_account_deal_agent_document

ChatGPT
Save a document into the deal plan get_account_deal_agent_results reads — an Account Plan, Mutual Action Plan or custom Deal Agent — as a NEW version. Use it to send a plan you wrote here into Aircover. Call get_account_deal_agent_results first: its declared_sections gives the section ids to key sections by, and its current_version is what to pass as base_version (required once the document has a version). Section ids are ids, never headings. A section you omit keeps the text it has; a section you send as an empty string is cleared; nothing else changes. Content that fits no declared section goes at the END of the LAST section's markdown under its own ### heading named after what it is, with that heading passed as leftover_heading — otherwise it is dropped. History is append-only, so this never overwrites the previous version — a rep can restore it from the plan's version dropdown in the web app. Sending the same document twice saves nothing the second time — that holds even when the base_version you hold has since moved on, so retrying a save you are not sure landed is safe and comes back as saved: false. The result carries a url: where the saved document now lives in Aircover, to hand to the user. For an ACCOUNT-level agent (account_plan, mutual_action_plan) that is the account view; for a deal_agent it is the deal view. Either way the agent's own tab is already open, so it is the link to give rather than one to assemble. Requires permission to edit that deal's plan: the deal's owner, someone on the deal team, or a manager/admin/owner.

share_account_deal_agent_document

ChatGPT
Share a deal plan — an Account Plan, Mutual Action Plan or custom Deal Agent — with teammates. THIS SENDS EMAIL: each person you name who does not already have access is emailed a link to the plan straight away, and from then on they are also emailed whenever the plan is updated by a run. So name only the people the user actually asked for, and tell the user who will be emailed. Everyone named must already be a member of the user's Aircover organization; an address that is not is refused and nobody is shared with. Sharing CANNOT be undone: there is no un-share, here or in the Aircover web app, so a person added to a plan stays on it. Someone who already has access is not emailed again, so repeating a call is safe and comes back as shared: false. Sharing grants READ access; it does not let a sharee edit the plan. Requires permission to manage that deal's plan: the deal's owner, someone on the deal team, or a manager/admin/owner.

share_report

ChatGPT
Share an intelligence report — the research list_reports and read_report read — with people at the user's own company. THIS SENDS EMAIL: each person you name who does not already have access is emailed a link to the report straight away. So name only the people the user actually asked for, and tell the user who will be emailed. Everyone named must have an email address at one of the user's own organization's domains; an address that is not is refused and NOBODY is shared with. They do NOT have to be an existing Aircover user. Sharing grants READ access, and somebody you share with can share the report on again. Somebody who already has access is not emailed again, so repeating a call is safe and comes back as shared: false. Use unshare_report to take access away again. Requires access to the report: its owner, or somebody it has already been shared with.

unshare_report

ChatGPT
Remove one person's access to an intelligence report, and drop the report from their list. Nobody is notified: no mail is sent, so tell the user that the person will simply stop seeing the report rather than hearing about it. The report's OWNER can remove anybody. Anybody else can only remove people they shared with themselves — you cannot undo somebody else's share, or remove your own access. Removing somebody who does not have access changes nothing and comes back as unshared: false, so repeating a call is safe. Takes ONE address per call. Requires access to the report: its owner, or somebody it has already been shared with.

agent_results

Claude

ask_documents

Claude

get_account_deal_agent_results

Claude

get_deal

Claude

get_deals

Claude

get_meeting

Claude

get_meeting_transcript

Claude

get_qualification_results

Claude

get_teams

Claude

list_account_deal_agents

Claude

list_agents

Claude

list_meetings

Claude

list_reports

Claude

read_report

Claude

save_account_deal_agent_document

Claude

share_account_deal_agent_document

Claude

share_report

Claude

unshare_report

Claude

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

App Stats

36

Tools

3

Prompts

ChatGPT, Claude

Platforms

Works with

ChatGPT
Claude

Data refreshed daily