pipeline import with CSV jinja issue

I have a pipeline like:

1.  Webhook: incoming request
   1. json body with 
      {

          "bucket": "some_bucket",

          "key": "processed/some_data.csv",

          "partner": "c1",

          "table_id": "abcdefghi",

          "headers": "a,b,c"

      }
2. S3:  Lookup an Object
   1. bucket = {{a.body_json.bucket}}
   2. key = {{a.body_json.key}}
3. Quickbase: import with CSV
   1. table = {{a.body_json.table_id}}
   2. merge field = fixed_merge_field
   3. csv url = {{b.file_transfer_handle}}
   4. first row is headers = True
   5. header row = {{a.body_params.headers}}

First two steps appear to work.  Third step does not expand the jinja template and appears to insert a literal string.  Is jinja prohibited in certain fields?  Do I need some extra magic to make this work?