Amazon DynamoDB
Register a DynamoDB source and query its items as flat, SQL-shaped tables.
DynamoDB stores documents, not rows. ChatPlotDB registers a connection, reads your items, and flattens their nested attributes into tables you can query the same way you query PostgreSQL.
Register a connection
Supply the region and credentials for the tables you want exposed. As with SQL connections, credentials are encrypted at rest.
How items become tables
Nested maps and lists are flattened into columns. A document like:
{ "id": "a1", "customer": { "name": "Acme", "tier": "pro" }, "total": 420 }becomes columns id, customer.name, customer.tier, and total. Querying
then works through DuckDB rather than against DynamoDB directly.
Refreshing
DynamoDB sources are snapshotted rather than queried live, so a refresh is an explicit step:
POST /api/configs/dynamodb/{key}/refreshSnapshots go stale
Unlike a SQL connection, answers reflect the last refresh, not this second. If a dashboard card reads DynamoDB, check its "data as of" before trusting it in a meeting.
Schema drift
Document stores let two items disagree about their shape. Attributes that appear in only some items become sparse columns. If answers look wrong, the usual cause is a newer item shape that the last refresh did not see.