fix(rag): lift chat output cap to 32768 tokens — long answers no longer cut off

tests/unit/test_config.py

@@ -34,6 +34,8 @@ def test_defaults_match_locked_decisions(monkeypatch: pytest.MonkeyPatch) -> Non
     assert s.hybrid_lexical_candidates >= 1
     assert s.rrf_k >= 1
     assert s.top_n_docs >= 1
+    # Owner instruction 2026-08-22: answers may run up to 32 768 tokens.
+    assert s.max_output_tokens == 32_768
     assert len(s.suggestions) >= 3
     # A9 (revised): the import scope covers the seven A9 formats.
     assert s.import_extension_set == {
@@ -49,6 +51,12 @@ def test_env_override(monkeypatch) -> None:
     assert s.llm_chat_model == "juggernaut"
 
 
+def test_max_output_tokens_env_override(monkeypatch) -> None:
+    monkeypatch.setenv("BOR_MAX_OUTPUT_TOKENS", "1234")
+    s = _settings()
+    assert s.max_output_tokens == 1234
+
+
 def test_import_extensions_env_override_is_a_csv_list(monkeypatch) -> None:
     monkeypatch.setenv("BOR_IMPORT_EXTENSIONS", "md,yml")
     s = _settings()