fix(F-15): address Opus review findings (CRITICAL + WARN)

CRITICAL:
- cli.py: invert cite-boost formula: divide distance by (1 + alpha*pr)
  instead of multiply — high-PageRank papers now correctly rank higher
- cli.py: wire graph_min_corpus_size config into graph report; warning
  emitted to stderr so JSON stdout stays parseable

WARN:
- graph.py: document that damping is ignored in small-graph uniform branch
- cli.py: sort dangling citations before slicing for stable output
- cli.py: replace raise typer.Exit(0) with return in empty-graph path
- tests: fix *extra_args helper signatures; add cite-boost ordering test
  and small-corpus warning test

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

codex/cli.py

@@ -103,7 +103,8 @@ def search_paper(
             results = []
             for row in rows:
                 pr_score = pr.get(row["id"], 0.0)
-                boosted = row["distance"] * (1.0 + alpha * pr_score)
+                # distance is lower=better; divide to reduce distance for high-PR papers
+                boosted = row["distance"] / (1.0 + alpha * pr_score)
                 results.append((boosted, row))
             results.sort(key=lambda x: x[0])
             for boosted_dist, row in results:
@@ -535,20 +536,23 @@ def graph_report(
     output_json: bool = typer.Option(False, "--json", help="Output as JSON."),
 ) -> None:
     """Show citation graph report: top hub papers, dangling citations, cluster summary."""
+    from codex.config import get_settings
     from codex.db import get_conn
     from codex.graph import build_citation_graph, citation_pagerank, dangling_citations
 
+    settings = get_settings()
     with get_conn() as conn:
         graph = build_citation_graph(conn)
         known_ids = {row["id"] for row in conn.execute("SELECT id FROM papers").fetchall()}
 
     if graph.number_of_nodes() == 0:
         typer.echo("Citation graph is empty — ingest some papers first.")
-        raise typer.Exit(0)
+        return
 
+    n_papers = len(known_ids)
     pr = citation_pagerank(graph)
     hubs = sorted(pr.items(), key=lambda x: -x[1])[:top_n]
-    dangling = dangling_citations(graph, known_ids)
+    dangling = sorted(dangling_citations(graph, known_ids))
 
     if output_json:
         typer.echo(
@@ -565,6 +569,12 @@ def graph_report(
         )
         return
 
+    if n_papers < settings.graph_min_corpus_size:
+        typer.echo(
+            f"Warning: only {n_papers} ingested papers "
+            f"(recommended ≥ {settings.graph_min_corpus_size} for meaningful ranking).",
+            err=True,
+        )
     typer.echo(f"Graph: {graph.number_of_nodes()} nodes, {graph.number_of_edges()} edges")
     typer.echo("")
     typer.echo(f"TOP-{top_n} HUB PAPERS (PageRank)")

codex/graph.py

@@ -64,9 +64,13 @@ def citation_pagerank(graph: nx.DiGraph, *, damping: float = 0.85) -> dict[str,
     -------
     ``{paper_id: score}`` dict.  Scores sum to approximately 1.0.
 
-    When ``graph`` has fewer than :data:`_MIN_PAGERANK_NODES` nodes,
-    returns a uniform distribution and logs a warning — the signal is
-    too thin to be meaningful.
+    When ``graph`` has fewer than :data:`_MIN_PAGERANK_NODES` (5) nodes,
+    returns a uniform distribution and logs a warning — the graph is too
+    sparse for the random-walk model to converge meaningfully.  The
+    ``damping`` parameter is ignored in this branch.  The user-facing
+    corpus-size threshold (default 15) lives in
+    :attr:`codex.config.Settings.graph_min_corpus_size` and is surfaced
+    by ``codex graph report``.
     """
     n = graph.number_of_nodes()
     if n == 0: