The task
Scenario. Atlan crawled 300 tables in the orders schema, and Context Agents Studio has already generated AI-ready descriptions and READMEs for all those assets. You also have extra context that lives elsewhere and could add to them: a spreadsheet of table owners, and a Confluence page of business knowledge to add. You want to pull that in without re-keying it by hand.
The ask
- The orders tables and what each still needs (owner, tags, extra description detail)
- Owners and tags applied from your spreadsheet
- Each description or README enriched with the details from Confluence
Hero prompt
I have a spreadsheet of table owners and a Confluence page describing the orders tables. Set the owners and tags from the spreadsheet, and enrich each table's description with the details from Confluence.
Step by step
-
Find the target assets and what each still needs.
Show the tables in the <schema> schema and flag what each still needs — owner, tags, extra description detail.
-
Match your outside context to them and show the plan.
Using my spreadsheet of owners/tags and the Confluence page on <schema>, match each row to its table and show what you'll set before writing.
-
Apply the owners, tags, and enriched descriptions.
Set the owners and tags, and add the Confluence details to each table's description or README.
What you'll give it
Required context
- The schema or keyword that scopes the assets (orders)
- Your outside context: a spreadsheet, a Confluence page, or similar
- Permission to update asset metadata
Optional context
- Which fields the external source maps to (which column is owner, which is tag)
- Descriptions and READMEs already generated by Context Agents Studio, so this only layers on any extra detail
- A house style for how the added detail should read
What Atlan creates
Outputs
- The orders assets matched to your spreadsheet / Confluence source, with the gaps shown
- Owners and tags set from the spreadsheet, previewed before anything is saved
- Descriptions and READMEs enriched with the Confluence details, on top of what Context Agents Studio already produced
Summary. The context you already have, sitting in a spreadsheet or a wiki, lands on the assets: owned, tagged, and described with detail Studio couldn't infer from the tables alone.
Changed: owners, tags, and description/README detail on the matched orders tables. Left untouched: the data, the schema, and the Context Agents Studio baseline you're adding to.
Follow-up prompts
Keep going once the first answer lands.
Extend to the next schema
Do the same for the payments schema, but skip any table that already has a description.
Certify the ready ones
Mark every table that now has an owner, tags, and a description as Verified.
Tag while you're in there
For the same tables, add a Domain: Orders tag and flag any column that looks like PII for me to review.
Tips & troubleshooting
Run it on one schema first. Review what it proposes before you let it write across the rest of your assets.
- Match the tool to the work. This writes metadata, so run it somewhere that shows you the proposed changes before it saves. Approve the batch, then let it write.
- Automate or chain it. Once you're happy with the results, point the same prompt at a whole database on a schedule, so newly crawled tables get AI-ready descriptions the day they land.
- Ask what it changed. Add "list every asset you changed and every field you wrote" to the prompt. That list is your record of what got enriched.