MCP tools - Perform.AI

All tools are read-only and operate on your selected organization. You don’t call tools directly, because your AI assistant chooses them, but knowing what exists helps you ask better questions.

Session

Tool	What it does
`whoami`	Shows the signed-in user and the organization the session acts on.
`list_organizations`	Lists the organizations your account belongs to.
`select_organization`	Selects the active organization for the session, by organization ID or slug (exactly one). All subsequent organization-scoped calls act on it. It changes only the session’s organization, not your data, though ChatGPT lists it under Write tools.
`list_resources`	Lists the workflow guides published by the server (below), optionally ranked by keywords. Works before an organization is selected.
`read_resource`	Reads one workflow guide by URI.

Shipment overview

Tool	What it does
`list_shipments`	Searches your organization’s Shipments using free text or structured filters, over a date window applied to a lifecycle date (imported, shipped, delivered, or issue date). Returns up to 10 rows per page. Field output follows a projection preset: `concise` (default, 16 fields), `basic` (39), or `verbose` (all), with per-field include/exclude.
`list_shipments_filters`	Returns the catalogue of filterable Shipment fields for your organization, with descriptions, data types, and grammar hints. The assistant calls this before building structured filters.
`get_shipment_filter_values`	Lists the actual values available for one filter field in your organization within a date window (your carrier references, your tags…), each with its Shipment count, which also answers “top N by shipment count” questions.
`list_master_data_event_types`	Searches the global event-type catalogue to resolve a human term (“delivered”, “in transit”) to the canonical key used in Shipment filters.
`list_shipments_bookmarks`	Lists your organization’s saved Portal bookmarks.
`list_shipments_from_saved`	Replays a saved Portal view (a bookmark, or the `fc` hash from a pasted Portal URL) as a live Shipment query, returning both the results and the exact filters applied.

Filter semantics: filters combine with AND across entries; multiple values within one entry combine with OR (a few fields support AND-within, indicated in the filter catalogue). When a free-text search is provided, structured filters in that request are ignored.

Shipment details

Tool	What it does
`get_shipment_details`	Opens one Shipment by exactly one identifier: Shipment UUID (globally unique), Shipment primary key, or your own shipment ID / customer reference. If a customer reference matches multiple Shipments across a parent-child organization tree, the first match is returned with an explicit disclosure so you can disambiguate by UUID. Supports the same `concise`/`basic`/`verbose` projections as `list_shipments`.

Carrier data

Tool	What it does
`list_carriers`	Searches Perform.AI’s master carrier catalogue: every carrier Perform.AI supports, independent of your organization. Filter by country, tracking integration type, retrievable data points (EDD, POD, pickup), booking type, and more. Answers “what carriers does Perform.AI support?”
`get_carrier`	The full master-catalogue record for one carrier, by its numeric ID.
`list_carriers_configs`	Lists the carriers configured for your organization (name, code, reference, status, outbound/return definition). Answers “which carriers do we use?”
`get_carrier_config`	Full detail for one of your organization’s carrier configurations, by its numeric ID. Config IDs belonging to other organizations return not-found.

Master catalogue vs your configs: list_carriers answers what Perform.AI supports globally; list_carriers_configs answers what your organization has set up. The same carrier can appear multiple times in your configs with different references (for example two DHL Express accounts). That’s normal, not a duplicate.

Additional coverage

Beyond the reference above, the toolset also covers bookings (listings, details, metrics), invoices (listings, details, statistics, exchange rates), notification configurations and triggers, performance rules and categories, tracking pages, checkout configuration (domains, variants), comply milestones, and integration connections. Your assistant discovers all tools automatically on connection; the detailed per-tool reference for these areas is on its way to this page. The master data behind Shipment filters is available as tools: list_master_data_countries, list_master_data_event_stages, list_master_data_event_phases, and list_master_data_event_types translate between human terms and the keys used in filters.

Tools in the ChatGPT plugin

The ChatGPT plugin connects to https://mcp.perform.ai/mcp-invoke. Instead of one tool per operation, this URL exposes eight tools, and the assistant reaches the operations on this page through them. The data, permissions, and organization scoping are the same as on https://mcp.perform.ai/mcp.

Tool	What it does
`list_organizations`	Lists the organizations your account belongs to.
`select_organization`	Selects the active organization for the session. ChatGPT lists it under Write tools because it changes the session’s organization; it doesn’t modify your data.
`whoami`	Shows the signed-in user and the organization the session acts on.
`list_resources`	Lists the workflow guides (below).
`read_resource`	Reads one workflow guide by URI.
`list_methods`	Finds the operations you can call, by keywords (for example `["carrier", "config"]`), by group (such as `shipments`, `carriers`, `bookings`, or `master_data`), or both. Returns up to 10 operations. Only lists operations your permissions allow, and hides organization-scoped operations until you select an organization.
`describe_methods`	Returns the input and response schemas for one to ten operations, by their exact names from `list_methods`.
`invoke_method_read`	Runs one read operation by its exact name, such as `list_shipments` or `get_carrier_config`, with that operation’s parameters.

A typical exchange in ChatGPT: the assistant calls list_methods to find the operation that answers your question, describe_methods to read its parameters, then invoke_method_read to run it. All operations reachable through invoke_method_read are read-only.

Resources

The server publishes two workflow guides as MCP resources. They tell the assistant which tools to call, and in what order, for common tasks. Both work before an organization is selected.

URI	Contents
`resource://perform-ai/guides/shipment-overview`	The order of calls that finds, filters, and reads the Shipments of the selected organization.
`resource://perform-ai/guides/master-data`	Which master-data tool answers a question about countries, event phases, event stages, or event types, and how the four differ.

​Session

​Shipment overview

​Shipment details

​Carrier data

​Additional coverage

​Tools in the ChatGPT plugin

​Resources

Session

Shipment overview

Shipment details

Carrier data

Additional coverage

Tools in the ChatGPT plugin

Resources