Merge the Apify datasets you pick, remove duplicates by any fields, filter, keep/drop/rename fields and keep only new items across runs. Works under limited permissions via the dataset picker. $0.25 per 1,000 output items.
README only. Live 0.1.2 (n11Xf9mmDgOBe4K6Q, published 11:59 AM America/New_York) still had "Pricing (proposal, not live)" and "Proposed price" (README sha 8b5bcd28…, identical to this 0.1.2 source). Those headings now say "Pricing" and "Price", and an apify-actor-start $0.00005 row was added (one per GB of memory; default memory is 1,024 MB per actor.json). No code, default, schema or event-price change.
2026-11-11: 0.1.2
Never-empty default dataset: count items actually written with push_data and decide on that counter. Do not use get_metadata().item_count, which can lag at 0 on the platform (Ranger, demo run AN4jZEGNdhkWtda8D and merge SuQAhYhZhOLEs2b0f, 2026-11-11) and was adding a false "No items were written" status row to every default-dataset run.
input-items-1000 proposed price raised from $0.001 to $0.0015. Ranger measured $0.000578 per 1,000 items read (run eScg3epTGSlGW9FJW). A 30% margin after Apify's 20% cut needs at least $0.00103. item-output stays $0.00025.
Tests: lagging dataset metadata (item_count = 0 after writes) must not emit a status row; every input_schema property must have a non-empty title and description; price tests use $0.0015.
2026-11-11: first version (version 0.1)
New Actor. Merges the datasets picked in datasetIds (resourcePicker, resourceType: dataset, READ) in the order picked, filters, dedupes by fields (dotted paths, optional case/space normalization, keep first or last), optional only-new-items across runs (named key-value store), keep/drop/rename fields, optional _sourceDatasetId, writes to the default dataset or appends to outputDatasetId (resourcePicker, READ and WRITE).
Input datasets and the output dataset are opened by ID only (never by name), so a mistyped value never creates an empty dataset; it gives one free status item and the other datasets still run.
Empty input runs a free built-in demo (8 items, 5 written). The default dataset is never left empty (a free summary or status item is written when needed).
Proposed pay-per-event events: item-output ($0.00025) and input-items-1000 ($0.001 per started 1,000 items read). The run stops at the user's maximum charge per run.