diff --git a/tests/test_repl_properties.py b/tests/test_repl_properties.py
new file mode 100644
index 0000000..a5bf579
--- /dev/null
+++ b/tests/test_repl_properties.py
@@ -0,0 +1,72 @@
+import json
+from io import StringIO
+
+from hypothesis import assume, given
+from hypothesis import strategies as st
+
+from context_compiler import create_engine
+from context_compiler.repl import run_repl
+
+LINE_TEXT = st.text(
+    alphabet=st.characters(blacklist_characters="\n\r"),
+    max_size=80,
+)
+
+
+def _run_repl_lines(lines: list[str]) -> tuple[str, list[dict[str, object]]]:
+    in_stream = StringIO("".join(f"{line}\n" for line in lines))
+    out_stream = StringIO()
+    run_repl(in_stream, out_stream)
+    raw = out_stream.getvalue()
+    parsed = [json.loads(line) for line in raw.splitlines() if line.strip()]
+    return raw, parsed
+
+
+@given(st.lists(LINE_TEXT, min_size=0, max_size=40))
+def test_repl_matches_engine_for_non_exit_sequences(lines: list[str]) -> None:
+    assume(all(line.strip().lower() not in {"exit", "quit"} for line in lines))
+
+    _, repl_decisions = _run_repl_lines(lines)
+
+    engine = create_engine()
+    oracle = [engine.step(line) for line in lines]
+    assert repl_decisions == oracle
+
+
+@given(st.lists(LINE_TEXT, min_size=0, max_size=40))
+def test_repl_is_deterministic_for_same_input(lines: list[str]) -> None:
+    raw1, parsed1 = _run_repl_lines(lines)
+    raw2, parsed2 = _run_repl_lines(lines)
+
+    assert raw1 == raw2
+    assert parsed1 == parsed2
+
+
+@given(
+    st.lists(LINE_TEXT, min_size=0, max_size=20),
+    st.sampled_from(["exit", "quit", " EXIT ", "QuIt"]),
+    st.lists(LINE_TEXT, min_size=0, max_size=20),
+)
+def test_repl_stops_processing_after_exit_or_quit(
+    prefix: list[str], stop_token: str, suffix: list[str]
+) -> None:
+    assume(all(line.strip().lower() not in {"exit", "quit"} for line in prefix))
+    lines = [*prefix, stop_token, *suffix]
+    _, repl_decisions = _run_repl_lines(lines)
+
+    engine = create_engine()
+    oracle = [engine.step(line) for line in prefix]
+    assert repl_decisions == oracle
+
+
+@given(st.lists(LINE_TEXT, min_size=0, max_size=40))
+def test_repl_emits_canonical_json_per_output_line(lines: list[str]) -> None:
+    assume(all(line.strip().lower() not in {"exit", "quit"} for line in lines))
+    raw, _ = _run_repl_lines(lines)
+
+    for output_line in raw.splitlines():
+        if not output_line.strip():
+            continue
+        parsed = json.loads(output_line)
+        canonical = json.dumps(parsed, sort_keys=True, separators=(",", ":"))
+        assert output_line == canonical