feed_summary

Docket summaries

A pre-computed rollup with one row per docket, joining docket metadata to comment counts and comment-period dates. Powers feed/timeline views without scanning the full comments dataset. Built by build_feed_summary, sorted by modify_date descending.

Coverage. Derived. One row per docket over dockets, documents and comments; it covers exactly what those cover. (measured 2026-09-06)

• Parquet file: feed_summary.parquet
• MCP query_sql support: Configured; requires an available artifact.
• Publication status: Not established by this schema page or its measurement date.
• Row count: Not stated here; the MCP describe_table reply gives the live count under publication.

Column	Type	Description
docket_id	VARCHAR	Docket identifier. One row per docket.
agency_code	VARCHAR	Posting agency's short code.
title	VARCHAR	Docket title (from dockets).
docket_type	VARCHAR	Docket category (from dockets).
modify_date	VARCHAR	Docket last-modified timestamp; the table is sorted by this descending.
abstract	VARCHAR	Docket abstract (from dockets). Often null.
comment_count	BIGINT	Posted comments on the docket, summed from comments_index, less those on a posting Regulations.gov removed (documents.publisher_status). 0 when none. These are posted comment records, not comments received: an agency posts one record for a mass-mail campaign that can stand for thousands of submissions (EPA-HQ-OW-2022-0114 has 1,629 posted records for about 122,200 comments EPA received). To estimate submissions received, sum comments.duplicate_comments as its description says; this table does not carry it.
comment_end_date	VARCHAR	Latest comment-period end date across the docket's documents, less postings Regulations.gov removed. Null when unknown.
date_created	VARCHAR	Earliest document posted_date on the docket, less postings Regulations.gov removed, used as a creation date. Null when unknown.