Step by step
Choose the atom CSV format
Open Create, choose Atom creation, then select CSV import. Use Classic CSV for image rich atoms on Mainnet or Testnet. Choose Unchained types Early access when you need a structured type file on Testnet.
Download and populate the right sample
Download the sample that matches the atoms you want to create. Keep its header row, then replace every example row with the real names, descriptions, URLs, image URLs, types, and other values you want written to the knowledge graph. Use one atom per row and delete any example rows you do not need.
Your spreadsheet should contain the actual atom data you want Collate to create, not instructions, sample content, or placeholder values. For a large dataset, download the sample first, then give that file and your source data to an AI assistant to populate the rows faster. Check every field before uploading and do not let it invent facts, links, or images.
Using Unchained Early access? Choose a type in CSV import to download its matching sample.
Save, upload, or paste the CSV
Export the completed spreadsheet as a CSV file, then select Upload CSV file. You can also paste the complete CSV text into the large input. Uploading only loads the data into Collate; it does not publish anything.
Preview exactly what Collate parsed
Select Preview CSV rows. Compare every parsed name, type, field, and Classic image preview with your spreadsheet. If a row shows an error or the values appear in the wrong columns, correct the source file and preview it again.
Review the atoms against the graph
Select Review atoms. Collate validates required fields, checks for atoms that already exist, detects repeated rows in the file, and surfaces same name matches that need attention.
Confirm exactly what is eligible
Read every review status and confirm the eligible count. Only Ready to create rows enter the transaction. Existing, blocked duplicate, ambiguous, and invalid rows remain excluded.
Publish the reviewed CSV batch
Connect your wallet on the selected network and select Publish eligible CSV atoms for Classic or Publish eligible atoms for Unchained. Approve one transaction and wait for confirmation before changing the file or leaving the page.
Example CSV
name,description,url,image_url,deposit
Knowledge Garden,A shared place for ideas,https://example.com,,0Keep in mind
- • Keep every header unique and leave optional cells blank rather than deleting required columns.
- • Start with a small Testnet file when using a new template for the first time.
What review may show
Ready to create
Parsed and eligible to publish.
Existing
Already exists and stays out.
Blocked duplicate
Repeated in the imported file.
Invalid
The row explains what must be corrected.