In my shop, we completed installing OEM agents on all the nodes to be monitored a few months back. However, the MGMT_TABLESPACE continues to grow. I figured that could be reasonable since I didn’t know the details of data retention parameters that govern when data gets purged. If a purge parameter was set to 6 months, but we had only reached our high number of targets 3 months ago, then naturally the database would keep growing.

I found these two links that explain purge parameters nicely.
How to change retention parameters for metrics history
How to change retention parameters for job history

There are three tables that contain the bulk of the data in the OEM tablespace MGMT_TABLESPACE. They are MGMT_METRICS_RAW, MGMT_METRICS_1HOUR and MGMT_METRICS_1DAY. They contain partitions that are named after time intervals, and OEM takes care of adding and dropping them. The default retention period for the raw data is 7 days, for the 1hour aggregate is 31 days and for the 1day aggregate is 365 days.

If you want to change the 7 day default period for MGMT_METRICS_RAW to 3 days, you must insert a record into the MGMT_PARAMETERS table to add the parameter mgmt_raw_keep_window.
The companion parameters for the other two tables are mgmt_hour_keep_window and mgmt_day_keep_window.