MCP App Store
by KeenEthics
Perform.AI icon

Perform.AI MCP Server for Claude and ChatGPT

By Perform.AI

Overview

Perform AI MCP gives your AI assistant direct access to your delivery and post-purchase data - the same data behind our web portal, available conversationally. The connector provides capabilities such as: segment shipments across your full filter catalog, drill into any parcel, replay a saved portal view as a live query, and check carrier coverage and configuration. The connector expands as the platform does, across post-purchase, checkout, returns, and logistics. Every capability your team reaches for on the portal should be reachable by an agent here, bounded by the same permission controls.

How to connect

  1. Claude: Open the Perform.AI listing in the Claude connectors directory and choose Connect. Claude listing Perform.AI exposes 8 tools here.
  2. ChatGPT: Open the Perform.AI listing in the ChatGPT app directory and choose Connect. ChatGPT listing Perform.AI exposes 8 tools here.

Tools

describe_methods

ChatGPT
Describe SDK methods by name Describe one to ten SDK methods by their exact names from list_methods. Each description comes with the JSON schema of the method input and the JSON schema of its response. A name that matches no method the caller may call is left out, and the message names it. Load the schema of invoke_method_read together with the schemas of list_methods and describe_methods, so that invoke_method_read is ready before the first call.

invoke_method_read

ChatGPT
Call one read SDK method by name Invoke a method of intent read by its exact snake_case name from list_methods. Call list_methods first to get valid names and the intent of each. Do not guess or use camelCase. Call describe_methods to see each method's parameter schema. The answer carries the response code, message and data of the called method. A method of intent write or destructive does not run here: it returns InvalidInput and the message names invoke_method_write. An unknown method returns NotFound. Parameters that fail validation return InvalidInput.

list_methods

ChatGPT
Find the SDK methods the caller may call Find the SDK methods that the caller may call, by keywords, by groups, or by both. Set `keywords`, `groups`, or both. Keywords are 1 to 4 nouns that name the data, for example ["carrier", "config"]. With keywords, the tool ranks the methods by relevance, best first. With groups and no keywords, the tool lists the methods in those groups by name. intent only narrows a call: it takes read, write or destructive, and the call still needs keywords, groups, or both. Each entry names its intent. A read method runs in invoke_method_read. A write or destructive method runs in invoke_method_write. The result holds at most 10 methods. If no method matches, the response code is NotFound, and the message names nouns to try. The methods run only inside execute_code, invoke_method_read or invoke_method_write, never as MCP tools. Pass the exact name. Do not guess method names. Load the schema of invoke_method_read together with the schemas of list_methods and describe_methods, so that invoke_method_read is ready before the first call. Workflow guides give the order of calls for a multi-step task: list_resources lists them, and read_resource reads one. GROUPS: bookings, carriers, checkout_widget, integrations, invoices, master_data, notifications, performance, shipments, tracking_pages.

list_organizations

ChatGPT
List the organizations the caller can select Calls no upstream — resolves against the caller's own validated membership.

list_resources

ChatGPT
List the workflow guides, or rank them by keywords List the workflow guides. Each guide gives the order of calls for one multi-step task. Works before an organization is selected. With no keywords, the tool returns every guide, sorted by URI. With keywords, the tool returns the guides that match, best first. Pass a returned URI to read_resource to read the guide.

read_resource

ChatGPT
Read one workflow guide by its exact URI Read one workflow guide by the exact URI that list_resources returns. Works before an organization is selected. When the account cannot use some tools that the guide names, the message names them. A well-formed URI that matches no guide exactly returns NotFound.

select_organization

ChatGPT
Select the organization a session acts on Calls no upstream — the selection is stored server-side against the session.

whoami

ChatGPT
Show who the session acts as Show the signed-in user and the organization the session acts on. The organization carries the permission scopes that decide which tools the session can use. The user is shown before and after an organization is selected. The organization is null until one is selected. When the user cannot be read, the call fails with a message that says why.

describe_methods

Claude

invoke_method_read

Claude

list_methods

Claude

list_organizations

Claude

list_resources

Claude

read_resource

Claude

select_organization

Claude

whoami

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/perform-ai"><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/perform-ai)