Match an invoice line, check stock, issue a purchase order, rate a shipment, open an exception. The agent picks the tool, calls it with typed parameters, and the log records the records read and the cost.
Agents call the first three. The fourth is how your own AI tools reach CommerceShip data.
Built-in tools
Rate a shipment, match an invoice line, check stock, allocate an order, compare cost across carriers. CommerceShip exposes its own operations to agents with typed parameters.
Data-scoped tools
A tool limited to the tables, fields, and rows a skill may see. Set the allow and deny lists and the row filters on the skill, and the tool cannot read past that scope.
Webhook tools
Bind a skill to a workflow or a webhook path, so an agent can act in a system CommerceShip does not connect to directly.
Your AI tools
Connect your own AI tools to CommerceShip data through the MCP server, with an API key on each request.
Watch the agent call them, then read them back on the record.
The agent calls the rating tool across your carriers, applies the rule that picks the service, then calls the label tool with the result. Each call carries its parameters and its answer.
Open a decision to see what the agent called, with which parameters, what came back, how long it took, and what it cost.
Agents call tools in all six Operations modules.
Pick a job from any module: an invoice to audit, a purchase order to issue, stock to check across locations. We'll show you which tools the agent calls and what the log records.
