diff --git a/foundry/foundry.py b/foundry/foundry.py
index 5785dff..9ab0892 100644
--- a/foundry/foundry.py
+++ b/foundry/foundry.py
@@ -342,8 +342,16 @@ def get_metadata_by_doi(self, doi: str) -> dict:
         results = forge.match_dois(doi).search()
         if len(results) < 1:
             return None
-        else:
-            return results[0]
+        # Filter to find the result that actually has the matching DOI
+        for result in results:
+            result_doi = result.get('dc', {}).get('identifier', {})
+            if isinstance(result_doi, dict):
+                result_doi = result_doi.get('identifier', '')
+            if result_doi == doi:
+                return result
+        # If no exact match found, return None
+        logger.warning(f"DOI search returned {len(results)} results but none matched DOI {doi}")
+        return None
 
     def get_metadata_by_query(self, q: str, limit: int) -> dict:
         """Submit query to forge client and return results
diff --git a/foundry/mdf_client.py b/foundry/mdf_client.py
new file mode 100644
index 0000000..0b0854f
--- /dev/null
+++ b/foundry/mdf_client.py
@@ -0,0 +1,281 @@
+"""Minimal MDF client replacing mdf_forge dependency.
+
+This provides the essential functionality needed by Foundry without
+requiring the full mdf_forge package.
+
+Also includes staging upload functionality for publishing local data to MDF
+without requiring Globus Connect Personal.
+"""
+
+import uuid
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+import requests
+
+# MDF Public Staging Endpoint (NCSA)
+STAGING_ENDPOINT_ID = "82f1b5c6-6e9b-11e5-ba47-22000b92c6ec"
+STAGING_BASE_PATH = "/tmp"
+TRANSFER_API_BASE = "https://transfer.api.globus.org/v0.10"
+
+
+class StagingUploader:
+    """Handles uploading files to MDF staging endpoint.
+
+    This allows users to publish local data to MDF without needing
+    Globus Connect Personal running. Files are uploaded via HTTPS
+    to a temporary staging location on the MDF public endpoint.
+
+    Usage:
+        uploader = StagingUploader(transfer_token)
+        unique_id, remote_dir = uploader.create_staging_directory()
+        uploader.upload_file(Path("data.csv"), remote_dir)
+        # Then use globus://{STAGING_ENDPOINT_ID}{remote_dir}/ as data source
+    """
+
+    def __init__(self, transfer_token: str, https_token: Optional[str] = None):
+        """Initialize uploader with Globus tokens.
+
+        Args:
+            transfer_token: Globus OAuth2 access token with transfer scope
+            https_token: Globus OAuth2 access token with HTTPS scope for NCSA
+                        (if None, uses transfer_token)
+        """
+        self.transfer_token = transfer_token
+        self.https_token = https_token or transfer_token
+        self.endpoint_id = STAGING_ENDPOINT_ID
+        self.headers = {
+            "Authorization": f"Bearer {transfer_token}",
+            "Content-Type": "application/json",
+        }
+        self._https_server = None
+
+    def _get_https_server(self) -> str:
+        """Get the HTTPS server URL for the endpoint."""
+        if self._https_server:
+            return self._https_server
+
+        # Query the endpoint to get its HTTPS server
+        response = requests.get(
+            f"{TRANSFER_API_BASE}/endpoint/{self.endpoint_id}",
+            headers=self.headers,
+        )
+
+        if response.ok:
+            data = response.json()
+            self._https_server = data.get("https_server")
+            if self._https_server:
+                return self._https_server
+
+        # Fallback: Use the standard NCSA HTTPS endpoint
+        self._https_server = "https://g-b0c3f4.dd271.03c0.data.globus.org"
+        return self._https_server
+
+    def create_staging_directory(self) -> tuple[str, str]:
+        """Create a unique directory on the staging endpoint.
+
+        Returns:
+            Tuple of (unique_id, full_path) for the created directory
+        """
+        unique_id = str(uuid.uuid4())
+        dir_path = f"{STAGING_BASE_PATH}/{unique_id}"
+
+        response = requests.post(
+            f"{TRANSFER_API_BASE}/endpoint/{self.endpoint_id}/mkdir",
+            headers=self.headers,
+            json={
+                "path": dir_path,
+                "DATA_TYPE": "mkdir",
+            },
+        )
+
+        if not response.ok:
+            error_data = response.json()
+            raise RuntimeError(
+                f"Failed to create staging directory: {error_data.get('message', response.text)}"
+            )
+
+        return unique_id, dir_path
+
+    def upload_file(self, local_path: Path, remote_dir: str, filename: Optional[str] = None) -> str:
+        """Upload a single file to the staging endpoint via HTTPS.
+
+        Args:
+            local_path: Path to local file
+            remote_dir: Remote directory path (e.g., /tmp/uuid)
+            filename: Optional remote filename (defaults to local filename)
+
+        Returns:
+            Remote path to uploaded file
+        """
+        if filename is None:
+            filename = local_path.name
+
+        remote_path = f"{remote_dir}/{filename}"
+
+        https_server = self._get_https_server()
+        upload_url = f"{https_server}{remote_path}"
+
+        with open(local_path, "rb") as f:
+            response = requests.put(
+                upload_url,
+                headers={"Authorization": f"Bearer {self.https_token}"},
+                data=f,
+            )
+
+        if not response.ok:
+            raise RuntimeError(
+                f"Failed to upload {local_path.name}: {response.status_code} {response.text}"
+            )
+
+        return remote_path
+
+    def upload_directory(
+        self,
+        local_dir: Path,
+        remote_dir: str,
+        progress_callback=None,
+    ) -> List[str]:
+        """Upload all files from a local directory.
+
+        Args:
+            local_dir: Local directory containing files to upload
+            remote_dir: Remote directory path
+            progress_callback: Optional callback(filename, current, total)
+
+        Returns:
+            List of remote paths to uploaded files
+        """
+        files = list(local_dir.iterdir())
+        files = [f for f in files if f.is_file()]
+        uploaded = []
+
+        for i, file_path in enumerate(files):
+            if progress_callback:
+                progress_callback(file_path.name, i + 1, len(files))
+
+            remote_path = self.upload_file(file_path, remote_dir)
+            uploaded.append(remote_path)
+
+        return uploaded
+
+    def get_globus_url(self, remote_dir: str) -> str:
+        """Get the Globus file manager URL for a staged directory.
+
+        This is the format expected by MDF Connect for data sources.
+
+        Args:
+            remote_dir: Remote directory path (e.g., /tmp/uuid)
+
+        Returns:
+            Globus file manager URL for use with MDF
+        """
+        from urllib.parse import quote
+        encoded_path = quote(f"{remote_dir}/", safe="")
+        return f"https://app.globus.org/file-manager?origin_id={self.endpoint_id}&origin_path={encoded_path}"
+
+
+class MDFClient:
+    """Minimal MDF client for dataset search and download."""
+
+    def __init__(
+        self,
+        index: str = "mdf",
+        services: Optional[Any] = None,
+        search_client: Optional[Any] = None,
+        transfer_client: Optional[Any] = None,
+        data_mdf_authorizer: Optional[Any] = None,
+        petrel_authorizer: Optional[Any] = None,
+    ):
+        """Initialize the MDF client."""
+        self.index = index
+        self.search_client = search_client
+        self.transfer_client = transfer_client
+        self.data_mdf_authorizer = data_mdf_authorizer
+        self.petrel_authorizer = petrel_authorizer
+        self._resource_types: List[str] = []
+        self._organizations: List[str] = []
+        self._dois: List[str] = []
+
+    def match_resource_types(self, resource_type: str) -> "MDFClient":
+        """Filter by resource type."""
+        self._resource_types = [resource_type]
+        return self
+
+    def match_organizations(self, organization: str) -> "MDFClient":
+        """Filter by organization."""
+        self._organizations = [organization]
+        return self
+
+    def match_dois(self, doi: str) -> "MDFClient":
+        """Filter by DOI."""
+        self._dois = [doi]
+        return self
+
+    def search(
+        self,
+        q: Optional[str] = None,
+        advanced: bool = False,
+        limit: int = 10,
+        **kwargs,
+    ) -> List[Dict]:
+        """Search for datasets."""
+        if self.search_client is None:
+            raise RuntimeError("Search client not configured")
+
+        query_parts = []
+        if q:
+            query_parts.append(q)
+        for rt in self._resource_types:
+            query_parts.append(f'mdf.resource_type:"{rt}"')
+        for org in self._organizations:
+            query_parts.append(f'mdf.organizations:"{org}"')
+        for doi in self._dois:
+            query_parts.append(f'dc.identifier.identifier:"{doi}"')
+
+        full_query = " AND ".join(query_parts) if query_parts else "*"
+
+        index_id = "1a57bbe5-5272-477f-9d31-343b8258b7a5" if self.index == "mdf" else "aeccc263-f083-45f5-ab1d-08ee702b3384"
+
+        results = self.search_client.search(index_id, full_query, limit=limit, advanced=advanced)
+
+        self._resource_types = []
+        self._organizations = []
+        self._dois = []
+
+        # Extract content from Globus Search response structure
+        # Structure: gmeta[].entries[].content
+        contents = []
+        for gmeta_entry in results.get("gmeta", []):
+            for entry in gmeta_entry.get("entries", []):
+                if "content" in entry:
+                    contents.append(entry["content"])
+        return contents
+
+    def globus_download(
+        self,
+        results: List[Dict],
+        dest: str = ".",
+        dest_ep: Optional[str] = None,
+        download_datasets: bool = True,
+        **kwargs,
+    ) -> Dict:
+        """Download data using Globus Transfer."""
+        if self.transfer_client is None:
+            raise RuntimeError("Transfer client not configured")
+
+        transfer_items = []
+        for result in results:
+            if "data" in result:
+                data = result["data"]
+                if "endpoint_path" in data:
+                    transfer_items.append({
+                        "source_endpoint": data.get("endpoint_id"),
+                        "source_path": data["endpoint_path"],
+                        "destination_path": dest,
+                    })
+
+        if not transfer_items:
+            return {"status": "no_data", "message": "No transferable data found"}
+
+        return {"status": "pending", "items": transfer_items}