diff --git a/graphify/llm.py b/graphify/llm.py
index 36b2afc48..e64c0c6c7 100644
--- a/graphify/llm.py
+++ b/graphify/llm.py
@@ -11,7 +11,7 @@
 import re
 import sys
 import time
-from collections.abc import Callable
+from collections.abc import Callable, Sequence
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from dataclasses import dataclass, replace
 from pathlib import Path
@@ -462,7 +462,7 @@ def _wrap_untrusted(rel: str, content: str) -> str:
     )
 
 
-def _read_files(units: "list[Path | FileSlice]", root: Path) -> str:
+def _read_files(units: "Sequence[Path | FileSlice]", root: Path) -> str:
     """Return file/slice contents formatted for the extraction prompt.
 
     Each unit is wrapped in an <untrusted_source> delimiter block and known
@@ -558,7 +558,7 @@ def _is_vision_image(path: Path) -> bool:
 
 
 def _partition_semantic_files(
-    units: "list[Path | FileSlice]",
+    units: "Sequence[Path | FileSlice]",
 ) -> tuple["list[Path | FileSlice]", list[Path]]:
     """Split a chunk into (text-like units, raster-image files).
 
@@ -1290,7 +1290,7 @@ def _call_bedrock(model: str, user_message: str, max_tokens: int = 8192, *, deep
 
 
 def extract_files_direct(
-    files: list[Path],
+    files: Sequence[Path | FileSlice],
     backend: str | None = None,
     api_key: str | None = None,
     model: str | None = None,
@@ -1516,7 +1516,7 @@ def _looks_like_context_exceeded(exc: BaseException) -> bool:
 
 
 def _extract_with_adaptive_retry(
-    chunk: list[Path],
+    chunk: Sequence[Path | FileSlice],
     backend: str,
     api_key: str | None,
     model: str | None,
@@ -1690,7 +1690,7 @@ def _split_lone_slice() -> "tuple[FileSlice, FileSlice] | None":
 
 
 def extract_corpus_parallel(
-    files: list[Path],
+    files: Sequence[str | Path],
     backend: str = "kimi",
     api_key: str | None = None,
     model: str | None = None,
@@ -1743,11 +1743,11 @@ def extract_corpus_parallel(
     # Split oversized splittable documents into slices that cover the whole file
     # before packing, so content past _FILE_CHAR_CAP is extracted instead of
     # silently dropped (#1369). Files at/under the cap pass through unchanged.
-    files = expand_oversized_files(files, _FILE_CHAR_CAP)
+    units = expand_oversized_files(files, _FILE_CHAR_CAP)
     if token_budget is not None:
-        chunks = _pack_chunks_by_tokens(files, token_budget=token_budget)
+        chunks = _pack_chunks_by_tokens(units, token_budget=token_budget)
     else:
-        chunks = [files[i:i + chunk_size] for i in range(0, len(files), chunk_size)]
+        chunks = [units[i:i + chunk_size] for i in range(0, len(units), chunk_size)]
 
     merged: dict = {
         "nodes": [], "edges": [], "hyperedges": [],
@@ -1756,7 +1756,9 @@ def extract_corpus_parallel(
     }
     total = len(chunks)
 
-    def _run_one(idx: int, chunk: list[Path]) -> tuple[int, dict | None, Exception | None]:
+    def _run_one(
+        idx: int, chunk: Sequence[Path | FileSlice]
+    ) -> tuple[int, dict | None, Exception | None]:
         t0 = time.time()
         try:
             result = _extract_with_adaptive_retry(
diff --git a/tests/test_chunking.py b/tests/test_chunking.py
index 087464ab8..2f9e3dfc8 100644
--- a/tests/test_chunking.py
+++ b/tests/test_chunking.py
@@ -277,6 +277,33 @@ def record(chunk, **kwargs):
     assert chunks_seen[0] == 50
 
 
+def test_corpus_parallel_accepts_string_paths(tmp_path):
+    """String file paths should be normalised before slicing and chunking."""
+    from graphify.llm import extract_corpus_parallel
+
+    files = []
+    for i in range(2):
+        f = tmp_path / f"f{i}.py"
+        f.write_text("x = 1\n")
+        files.append(f)
+
+    chunks_seen = []
+
+    def record(chunk, **kwargs):
+        chunks_seen.append(chunk)
+        return _stub_chunk_result(len(chunk), len(chunks_seen))
+
+    with patch("graphify.llm.extract_files_direct", side_effect=record):
+        result = extract_corpus_parallel(
+            [str(f) for f in files], backend="kimi", max_concurrency=1
+        )
+
+    assert len(chunks_seen) == 1
+    assert chunks_seen[0] == files
+    assert all(isinstance(p, Path) for p in chunks_seen[0])
+    assert len(result["nodes"]) == 2
+
+
 # ---- Adaptive retry on truncation -------------------------------------------
 
 def _stub_with_finish(file_count: int, finish_reason: str = "stop") -> dict: