diff --git a/custom_nodes/research/__init__.py b/custom_nodes/research/__init__.py
index e97efb426..15300b2a3 100644
--- a/custom_nodes/research/__init__.py
+++ b/custom_nodes/research/__init__.py
@@ -11,7 +11,8 @@ class ResearchExtension(ComfyExtension):
         from custom_nodes.research.claim_extract import PaperClaimExtract
         from custom_nodes.research.evidence_assemble import ClaimEvidenceAssemble
         from custom_nodes.research.style_profile import StyleProfileExtract
-        return [PaperSearch, PaperClaimExtract, ClaimEvidenceAssemble, StyleProfileExtract]
+        from custom_nodes.research.reference_paper_select import ReferencePaperSelect
+        return [PaperSearch, PaperClaimExtract, ClaimEvidenceAssemble, StyleProfileExtract, ReferencePaperSelect]
 
 
 async def comfy_entrypoint() -> ComfyExtension:
diff --git a/custom_nodes/research/reference_paper_select.py b/custom_nodes/research/reference_paper_select.py
new file mode 100644
index 000000000..08f681302
--- /dev/null
+++ b/custom_nodes/research/reference_paper_select.py
@@ -0,0 +1,59 @@
+"""ReferencePaperSelect node - select papers from project for canvas use."""
+import json
+from typing_extensions import override
+
+from comfy_api.latest import ComfyNode, io
+
+
+class ReferencePaperSelect(io.ComfyNode):
+    """Select papers from a project to load onto the canvas for reference."""
+
+    @classmethod
+    def define_schema(cls) -> io.Schema:
+        return io.Schema(
+            node_id="ReferencePaperSelect",
+            display_name="Select Papers",
+            category="Research",
+            inputs=[
+                io.String.Input(
+                    "project_id",
+                    display_name="Project ID",
+                    default="",
+                ),
+                io.Int.Input(
+                    "max_papers",
+                    display_name="Max Papers",
+                    default=5,
+                    min=1,
+                    max=20,
+                    step=1,
+                ),
+            ],
+            outputs=[
+                io.String.Output(display_name="Selected Papers (JSON)"),
+            ],
+        )
+
+    @classmethod
+    def execute(cls, project_id: str, max_papers: int) -> io.NodeOutput:
+        if not project_id.strip():
+            return io.NodeOutput(selected_papers=json.dumps([]))
+
+        # Phase 1: Returns mock data
+        # Phase 2+ would query the actual project papers from DB
+        return io.NodeOutput(selected_papers=json.dumps([
+            {
+                "paper_id": "mock-id-1",
+                "title": "Sample Paper 1",
+                "authors": ["Author A", "Author B"],
+                "abstract": "This is a sample abstract for demonstration.",
+                "year": "2024",
+            },
+            {
+                "paper_id": "mock-id-2",
+                "title": "Sample Paper 2",
+                "authors": ["Author C"],
+                "abstract": "Another sample abstract.",
+                "year": "2023",
+            }
+        ][:max_papers], indent=2))