Lookups and imports are the steps that do a tool's work: a lookup reads records from an application, and an import creates, updates, or deletes them. A tool needs at least one of them before it can run or return anything. This article shows integrator.io users how to add a lookup or import between the tool's input and output, configure it, and choose what happens when a record fails. For what a tool is, see Celigo tools overview.
Important
Prerequisites: Configure tool input. A lookup or import maps from the input, so the input schema must exist first.
Lookups and imports in a tool
- A lookup retrieves data — for example, a customer record by email — and merges the result back into the record through results mapping. See Lookups.
- An import writes data — creating, updating, or deleting records — and merges the application's response back through response mapping. See Imports.
You can build a new step or reuse an existing lookup or import from your account; to change an existing one for the tool without affecting its other users, clone it first.
Add a lookup or import
- Go to AI studio, and then select Tools.
- Select the tool.
- Select Add step to the right of Tool input, and then select Add destination / lookup.
- Choose the application, or select a pre-built step, or describe the step in your own words to get a recommendation. For the connector list, see Connectors.
- Choose whether the step is a lookup or an import.
- Enter a Name and Description.
- Select an existing connection, or create one. See Create a connection.
- Build the step as you would in a flow. For an example, see Create an HTTP import.
The step appears on the canvas with its application's logo and type (Lookup or Import). To add another step, select Add step after it.
If the application returns XML or CSV, turn on Override request/response media type on the step. The tool's final output must still be JSON.
Configure the step
Select More on the step. Under Incoming data, Input filter processes only records that match a condition, and Mapping maps the record's fields into the application's fields. Under Outgoing data, Transformation, Output filter, Pre save page hook, Results mapping, Post response map hook, and Error handling act on what the application returns.
- Input and output filters
- Transformations
- Hooks and scripts
- Map data in a tool — the mapping types inside a tool
Choose what happens when a record fails
To choose what happens when a record fails, select More on the step > Error handling, and then choose what the failed record should do.
| Setting | Behavior | What the caller receives |
|---|---|---|
| Stop and return error (default) | The tool stops at this step. | An errors array with the code, message, and source of the failure, instead of the tool output. |
| Proceed to the next application regardless | The tool continues to the next step. | The tool output. The step's errors are available in its results or response mapping, so map the errors array into the output if the caller needs them. |
When a flow calls the tool, a returned error follows the flow's own error handling and retry. For a caller's view of errors, see Use a tool in flows, APIs, AI agents, and MCP servers.