diff --git a/ast_rag/cli.py b/ast_rag/cli.py
index e22741b..0d32f51 100644
--- a/ast_rag/cli.py
+++ b/ast_rag/cli.py
@@ -1806,5 +1806,116 @@ def cache_stats(
     console.rule("[bold green]Done[/bold green]")
 
 
+# ---------------------------------------------------------------------------
+
+
+def _collect_index_stats(driver) -> dict:
+    """Gather node/edge/language/file counts from the graph.
+
+    Edges are stored as generic ``:EDGE`` relationships carrying the semantic
+    type in a ``kind`` property (see ``repositories.queries.batch_upsert_edges``),
+    so edge types are grouped by that property rather than by relationship type.
+    """
+    stats: dict = {
+        "nodes": {"total": 0, "by_kind": {}},
+        "edges": {"total": 0, "by_kind": {}},
+        "languages": {},
+        "files": 0,
+    }
+
+    with driver.session() as session:
+        # CurrentVersion is the graph's bookkeeping pointer, not a code node.
+        for record in session.run(
+            "MATCH (n) WHERE n.id IS NOT NULL AND NOT n:CurrentVersion "
+            "RETURN labels(n)[0] AS kind, count(*) AS n ORDER BY n DESC"
+        ):
+            stats["nodes"]["by_kind"][record["kind"] or "unlabelled"] = record["n"]
+        stats["nodes"]["total"] = sum(stats["nodes"]["by_kind"].values())
+
+        for record in session.run(
+            "MATCH ()-[r]->() RETURN coalesce(r.kind, type(r)) AS kind, "
+            "count(*) AS n ORDER BY n DESC"
+        ):
+            stats["edges"]["by_kind"][record["kind"]] = record["n"]
+        stats["edges"]["total"] = sum(stats["edges"]["by_kind"].values())
+
+        for record in session.run(
+            "MATCH (n) WHERE n.lang IS NOT NULL "
+            "RETURN n.lang AS lang, count(*) AS n ORDER BY n DESC"
+        ):
+            stats["languages"][record["lang"]] = record["n"]
+
+        record = session.run(
+            "MATCH (n) WHERE n.file_path IS NOT NULL RETURN count(DISTINCT n.file_path) AS files"
+        ).single()
+        stats["files"] = record["files"] if record else 0
+
+    return stats
+
+
+@app.command("stats")
+def stats(
+    config: Optional[str] = typer.Option(None, "--config", "-c", help="Path to config JSON"),
+    as_json: bool = typer.Option(False, "--json", help="Emit JSON instead of a table"),
+    verbose: bool = typer.Option(False, "--verbose", "-v"),
+) -> None:
+    """
+    Show statistics about the indexed codebase.
+
+    Reports node counts by kind, edge counts by type, the language
+    distribution and the number of indexed files.
+
+    Examples:
+
+      ast-rag stats
+      ast-rag stats --json
+      ast-rag stats --config ast_rag_config.json
+    """
+    if verbose:
+        logging.basicConfig(level=logging.DEBUG)
+    else:
+        logging.basicConfig(level=logging.WARNING)
+
+    cfg = _load_config(config)
+    driver = create_driver(cfg.neo4j)
+    try:
+        data = _collect_index_stats(driver)
+    finally:
+        driver.close()
+
+    if as_json:
+        print(json.dumps(data, indent=2))
+        return
+
+    if data["nodes"]["total"] == 0:
+        console.print("[yellow]No indexed nodes found. Run `ast-rag init <path>` first.[/yellow]")
+        return
+
+    console.rule("[bold blue]AST-RAG Index Statistics[/bold blue]")
+    console.print(f"  Files indexed: {data['files']:,}")
+
+    console.print()
+    console.print(f"[bold]Nodes[/bold] ({data['nodes']['total']:,} total)")
+    for kind, count in data["nodes"]["by_kind"].items():
+        console.print(f"  {kind:<20} {count:>8,}")
+
+    console.print()
+    console.print(f"[bold]Edges[/bold] ({data['edges']['total']:,} total)")
+    if data["edges"]["by_kind"]:
+        for kind, count in data["edges"]["by_kind"].items():
+            console.print(f"  {kind:<20} {count:>8,}")
+    else:
+        console.print("  [dim]none[/dim]")
+
+    console.print()
+    console.print("[bold]Languages[/bold]")
+    if data["languages"]:
+        total = sum(data["languages"].values()) or 1
+        for lang, count in data["languages"].items():
+            console.print(f"  {lang:<20} {count:>8,}  ({count / total:.0%})")
+    else:
+        console.print("  [dim]none[/dim]")
+
+
 if __name__ == "__main__":
     app()
diff --git a/tests/test_stats_command.py b/tests/test_stats_command.py
new file mode 100644
index 0000000..10899e7
--- /dev/null
+++ b/tests/test_stats_command.py
@@ -0,0 +1,98 @@
+"""Tests for the `ast-rag stats` command (issue #13)."""
+
+from __future__ import annotations
+
+import json
+from unittest.mock import MagicMock
+
+from typer.testing import CliRunner
+
+from ast_rag.cli import _collect_index_stats, app
+
+runner = CliRunner()
+
+
+class _Rec(dict):
+    """Minimal stand-in for a neo4j Record."""
+
+    def __getitem__(self, key):
+        return super().__getitem__(key)
+
+
+def _driver_returning(node_rows, edge_rows, lang_rows, files):
+    """Build a mock driver whose session.run dispatches on the query text."""
+    session = MagicMock()
+
+    def run(query, **_kwargs):
+        result = MagicMock()
+        if "labels(n)[0]" in query:
+            return iter([_Rec(kind=k, n=n) for k, n in node_rows])
+        if "coalesce(r.kind" in query:
+            return iter([_Rec(kind=k, n=n) for k, n in edge_rows])
+        if "n.lang" in query:
+            return iter([_Rec(lang=lang, n=n) for lang, n in lang_rows])
+        result.single.return_value = _Rec(files=files)
+        return result
+
+    session.run.side_effect = run
+    driver = MagicMock()
+    driver.session.return_value.__enter__.return_value = session
+    return driver
+
+
+def test_collect_index_stats_shapes_and_totals():
+    driver = _driver_returning(
+        node_rows=[("Method", 368), ("Class", 100)],
+        edge_rows=[("CONTAINS_METHOD", 392), ("CALLS", 150)],
+        lang_rows=[("java", 468)],
+        files=103,
+    )
+    stats = _collect_index_stats(driver)
+
+    assert stats["files"] == 103
+    assert stats["nodes"]["total"] == 468
+    assert stats["nodes"]["by_kind"] == {"Method": 368, "Class": 100}
+    assert stats["edges"]["total"] == 542
+    assert stats["edges"]["by_kind"]["CALLS"] == 150
+    assert stats["languages"] == {"java": 468}
+
+
+def test_edges_grouped_by_kind_property_not_relationship_type():
+    """Edges are stored as :EDGE with the semantic type in `kind`."""
+    driver = _driver_returning([], [], [], 0)
+    _collect_index_stats(driver)
+    queries = [
+        c.args[0] for c in driver.session.return_value.__enter__.return_value.run.call_args_list
+    ]
+    edge_query = next(q for q in queries if "]->()" in q and "count(*)" in q)
+    assert "r.kind" in edge_query, "edge stats must group by the kind property"
+
+
+def test_bookkeeping_node_excluded():
+    driver = _driver_returning([], [], [], 0)
+    _collect_index_stats(driver)
+    queries = [
+        c.args[0] for c in driver.session.return_value.__enter__.return_value.run.call_args_list
+    ]
+    node_query = next(q for q in queries if "labels(n)[0]" in q)
+    assert "CurrentVersion" in node_query
+
+
+def test_empty_index_reports_hint(monkeypatch):
+    monkeypatch.setattr("ast_rag.cli.create_driver", lambda _cfg: _driver_returning([], [], [], 0))
+    result = runner.invoke(app, ["stats"])
+    assert result.exit_code == 0
+    assert "No indexed nodes found" in result.stdout
+
+
+def test_json_output_is_valid_json(monkeypatch):
+    monkeypatch.setattr(
+        "ast_rag.cli.create_driver",
+        lambda _cfg: _driver_returning([("Method", 5)], [("CALLS", 2)], [("python", 5)], 3),
+    )
+    result = runner.invoke(app, ["stats", "--json"])
+    assert result.exit_code == 0
+    payload = json.loads(result.stdout)
+    assert payload["nodes"]["by_kind"] == {"Method": 5}
+    assert payload["edges"]["by_kind"] == {"CALLS": 2}
+    assert payload["files"] == 3