diff --git a/.semversioner/next-release/patch-20240919000033794082.json b/.semversioner/next-release/patch-20240919000033794082.json
new file mode 100644
index 0000000000..522e6050c2
--- /dev/null
+++ b/.semversioner/next-release/patch-20240919000033794082.json
@@ -0,0 +1,4 @@
+{
+  "type": "patch",
+  "description": "Fix prompt tune output path"
+}
diff --git a/graphrag/prompt_tune/__main__.py b/graphrag/prompt_tune/__main__.py
index 1db893f2ff..aac668dba1 100644
--- a/graphrag/prompt_tune/__main__.py
+++ b/graphrag/prompt_tune/__main__.py
@@ -92,7 +92,7 @@
     )
     parser.add_argument(
         "--output",
-        help="Directory to save generated prompts to. Default: 'prompts'",
+        help="Directory to save generated prompts to, relative to the root directory. Default: 'prompts'",
         type=str,
         default="prompts",
     )
diff --git a/graphrag/prompt_tune/cli.py b/graphrag/prompt_tune/cli.py
index 0f4c795bf1..dd964260ca 100644
--- a/graphrag/prompt_tune/cli.py
+++ b/graphrag/prompt_tune/cli.py
@@ -43,7 +43,7 @@ async def prompt_tune(
     - chunk_size: The chunk token size to use.
     - language: The language to use for the prompts.
     - skip_entity_types: Skip generating entity types.
-    - output: The output folder to store the prompts.
+    - output: The output folder to store the prompts. Relative to the root directory.
     - n_subset_max: The number of text chunks to embed when using auto selection method.
     - k: The number of documents to select when using auto selection method.
     - min_examples_required: The minimum number of examples required for entity extraction prompts.
@@ -67,7 +67,7 @@ async def prompt_tune(
         k=k,
     )
 
-    output_path = Path(output).resolve()
+    output_path = (root_path / output).resolve()
     if output_path:
         reporter.info(f"Writing prompts to {output_path}")
         output_path.mkdir(parents=True, exist_ok=True)