Sparse-day-proof briefs, feed health check, and 16 new sources

- Briefs now fill from a rolling window (prefer today, backfill up to
  window_days) and exclude anything featured in the last 7 days of briefs, so
  slow days still produce five items without stories lingering day to day.
- New 'check-feeds' command fetches and parses every feed to catch dead ones.
- Added 16 validated sources (science, environment, animals, culture),
  expanding coverage from 12 to 28 feeds to reduce staleness.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

config/sources.toml

@@ -118,3 +118,147 @@ pr_risk_score = 4
 poll_interval_minutes = 120
 notes = "University research stories; watch PR framing."
 
+
+[[sources]]
+name = "Phys.org"
+feed_url = "https://phys.org/rss-feed/"
+default_category = "science"
+trust_score = 7
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Broad science aggregator."
+
+[[sources]]
+name = "Nature News"
+feed_url = "https://www.nature.com/nature.rss"
+default_category = "science"
+trust_score = 9
+pr_risk_score = 2
+poll_interval_minutes = 180
+notes = "Top-tier science journal news."
+
+[[sources]]
+name = "Ars Technica Science"
+feed_url = "https://feeds.arstechnica.com/arstechnica/science"
+default_category = "science"
+trust_score = 7
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Science/tech reporting."
+
+[[sources]]
+name = "The Guardian Science"
+feed_url = "https://www.theguardian.com/science/rss"
+default_category = "science"
+trust_score = 8
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Mainstream science; needs filtering."
+
+[[sources]]
+name = "The Guardian Environment"
+feed_url = "https://www.theguardian.com/environment/rss"
+default_category = "environment"
+trust_score = 8
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Environment; needs filtering."
+
+[[sources]]
+name = "ScienceDaily Plants & Animals"
+feed_url = "https://www.sciencedaily.com/rss/plants_animals.xml"
+default_category = "animals"
+trust_score = 6
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Animal/biology discoveries."
+
+[[sources]]
+name = "ScienceDaily Space & Time"
+feed_url = "https://www.sciencedaily.com/rss/space_time.xml"
+default_category = "science"
+trust_score = 6
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Space discoveries."
+
+[[sources]]
+name = "Smithsonian Magazine"
+feed_url = "https://www.smithsonianmag.com/rss/latest_articles/"
+default_category = "culture"
+trust_score = 7
+pr_risk_score = 3
+poll_interval_minutes = 180
+notes = "Culture, history, science human-interest."
+
+[[sources]]
+name = "Yale Environment 360"
+feed_url = "https://e360.yale.edu/feed.xml"
+default_category = "environment"
+trust_score = 8
+pr_risk_score = 2
+poll_interval_minutes = 180
+notes = "In-depth environment/solutions."
+
+[[sources]]
+name = "Anthropocene Magazine"
+feed_url = "https://www.anthropocenemagazine.org/feed/"
+default_category = "environment"
+trust_score = 7
+pr_risk_score = 3
+poll_interval_minutes = 180
+notes = "Sustainability solutions journalism."
+
+[[sources]]
+name = "The Conversation (US)"
+feed_url = "https://theconversation.com/us/articles.atom"
+default_category = "science"
+trust_score = 7
+pr_risk_score = 2
+poll_interval_minutes = 120
+notes = "Academic-authored explainers."
+
+[[sources]]
+name = "Colossal"
+feed_url = "https://www.thisiscolossal.com/feed/"
+default_category = "culture"
+trust_score = 6
+pr_risk_score = 3
+poll_interval_minutes = 240
+notes = "Art and visual culture."
+
+[[sources]]
+name = "Atlas Obscura"
+feed_url = "https://www.atlasobscura.com/feeds/latest"
+default_category = "culture"
+trust_score = 6
+pr_risk_score = 3
+poll_interval_minutes = 240
+notes = "Curiosities, places, culture."
+
+[[sources]]
+name = "New Scientist"
+feed_url = "https://www.newscientist.com/feed/home/"
+default_category = "science"
+trust_score = 7
+pr_risk_score = 3
+poll_interval_minutes = 120
+notes = "Science weekly."
+
+[[sources]]
+name = "ScienceAlert"
+feed_url = "https://www.sciencealert.com/feed"
+default_category = "science"
+trust_score = 6
+pr_risk_score = 4
+poll_interval_minutes = 120
+notes = "Pop science; watch hype."
+
+[[sources]]
+name = "Grist"
+feed_url = "https://grist.org/feed/"
+default_category = "environment"
+trust_score = 6
+pr_risk_score = 3
+poll_interval_minutes = 180
+notes = "Climate solutions reporting."

goodnews/briefs.py

@@ -9,6 +9,7 @@ def build_daily_brief(
     brief_date: str | None = None,
     limit: int = 5,
     replace: bool = False,
+    window_days: int = 3,
 ) -> int:
     target_date = brief_date or date.today().isoformat()
     existing = conn.execute("SELECT id FROM daily_briefs WHERE brief_date = ?", (target_date,)).fetchone()
@@ -22,7 +23,7 @@ def build_daily_brief(
         (target_date, f"Five Good Things Today - {target_date}"),
     ).lastrowid
 
-    rows = _candidate_articles(conn, target_date)
+    rows = _candidate_articles(conn, target_date, window_days)
     selected = _select_diverse(rows, limit)
     for index, row in enumerate(selected, start=1):
         conn.execute(
@@ -78,7 +79,17 @@ def show_brief(conn: sqlite3.Connection, brief_date: str | None = None, limit: i
     ).fetchall()
 
 
-def _candidate_articles(conn: sqlite3.Connection, target_date: str) -> list[sqlite3.Row]:
+def _candidate_articles(
+    conn: sqlite3.Connection, target_date: str, window_days: int = 3
+) -> list[sqlite3.Row]:
+    """Brief candidates, sparse-day-proof.
+
+    Prefers articles dated on target_date, but widens to the preceding
+    `window_days` so the brief still fills on slow news days. Anything already
+    featured in a brief within the last 7 days (other than this same date, which
+    is being rebuilt) is excluded so backfilled stories cannot linger across
+    consecutive days.
+    """
     return conn.execute(
         """
         SELECT
@@ -100,19 +111,31 @@ def _candidate_articles(conn: sqlite3.Connection, target_date: str) -> list[sqli
             s.pr_risk_score,
             s.reason_code,
             s.reason_text,
-            s.model_name
+            s.model_name,
+            CASE WHEN date(COALESCE(a.published_at, a.discovered_at)) = date(?)
+                 THEN 1 ELSE 0 END AS is_today
         FROM articles a
         JOIN sources src ON src.id = a.source_id
         JOIN article_scores s ON s.article_id = a.id
         WHERE s.accepted = 1
-          AND date(COALESCE(a.published_at, a.discovered_at)) = date(?)
+          AND date(COALESCE(a.published_at, a.discovered_at)) <= date(?)
+          AND date(COALESCE(a.published_at, a.discovered_at)) > date(?, '-' || ? || ' days')
+          AND a.id NOT IN (
+              SELECT bi.article_id
+              FROM daily_brief_items bi
+              JOIN daily_briefs b ON b.id = bi.brief_id
+              WHERE b.brief_date <> ?
+                AND b.brief_date <= date(?)
+                AND b.brief_date > date(?, '-7 days')
+          )
         ORDER BY
+            is_today DESC,
             (s.constructive_score + s.agency_score + s.human_benefit_score + src.trust_score
              - s.cortisol_score - s.ragebait_score - s.pr_risk_score) DESC,
             COALESCE(a.published_at, a.discovered_at) DESC
         LIMIT 50
         """,
-        (target_date,),
+        (target_date, target_date, target_date, window_days, target_date, target_date, target_date),
     ).fetchall()
 
 

goodnews/cli.py

@@ -8,7 +8,7 @@ from pathlib import Path
 
 from .briefs import build_daily_brief, show_brief
 from .db import connect, init_db
-from .feeds import poll_all_sources, poll_due_sources, poll_source
+from .feeds import fetch_feed, parse_feed, poll_all_sources, poll_due_sources, poll_source
 from .llm import LocalModelClient, classify_articles
 from .scoring import score_article
 from .sources import load_sources, upsert_sources
@@ -48,6 +48,9 @@ def main() -> None:
 
     subparsers.add_parser("source-report", help="Show source-level ingestion and scoring stats")
 
+    check_feeds_parser = subparsers.add_parser("check-feeds", help="Fetch and parse each feed, reporting health")
+    check_feeds_parser.add_argument("--all", action="store_true", help="Include inactive sources")
+
     runs_parser = subparsers.add_parser("list-runs", help="Show recent ingest runs")
     runs_parser.add_argument("--limit", type=int, default=20)
 
@@ -122,6 +125,8 @@ def main() -> None:
         list_category(conn, topic=args.topic, flavor=args.flavor, limit=args.limit, accepted_only=not args.all)
     elif args.command == "source-report":
         source_report(conn)
+    elif args.command == "check-feeds":
+        check_feeds(conn, include_inactive=args.all)
     elif args.command == "list-runs":
         list_runs(conn, limit=args.limit)
     elif args.command == "rescore":
@@ -214,6 +219,20 @@ def list_recent(conn: sqlite3.Connection, limit: int, accepted_only: bool) -> No
         print(f"  {row['canonical_url']}")
 
 
+def check_feeds(conn: sqlite3.Connection, include_inactive: bool = False) -> None:
+    where = "" if include_inactive else "WHERE active = 1"
+    rows = conn.execute(f"SELECT name, feed_url FROM sources {where} ORDER BY name").fetchall()
+    ok = 0
+    for row in rows:
+        try:
+            items = parse_feed(fetch_feed(row["feed_url"]))
+            ok += 1
+            print(f"OK   {row['name']}: {len(items)} items")
+        except Exception as exc:
+            print(f"FAIL {row['name']}: {exc}")
+    print(f"--- {ok}/{len(rows)} feeds healthy ---")
+
+
 def run_cycle(conn: sqlite3.Connection, args: argparse.Namespace) -> None:
     """One end-to-end pass for a scheduler: poll due sources, classify the new
     arrivals, rebuild today's brief. Each step is independent and non-fatal so a