Microsoft Graph — File Operations

Manage files on OneDrive and SharePoint document libraries using the Microsoft Graph MCP server. Use this for storing, retrieving, and organizing network documentation, configuration backups, audit reports, and diagram artifacts.

How to Call the Tools

The Microsoft Graph MCP server is invoked via npx with Azure AD credentials:

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" <tool-name> '<arguments-json>'

Available Operations

1. List Files in a Drive or Folder

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_list_files '{"driveId":"<drive-id>","folderId":"root"}'

2. Search for Files

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_search_files '{"query":"network topology diagram","driveId":"<drive-id>"}'

3. Download / Read File Content

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_get_file_content '{"driveId":"<drive-id>","itemId":"<item-id>"}'

4. Upload a File

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_upload_file '{"driveId":"<drive-id>","parentId":"<folder-id>","fileName":"health-report-2026-02-22.md","content":"<file-content>"}'

5. Create a Folder

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_create_folder '{"driveId":"<drive-id>","parentId":"root","folderName":"NetClaw Reports"}'

6. List SharePoint Sites

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_list_sites '{}'

7. Get Site Drive

AZURE_TENANT_ID=$AZURE_TENANT_ID AZURE_CLIENT_ID=$AZURE_CLIENT_ID AZURE_CLIENT_SECRET=$AZURE_CLIENT_SECRET \
  python3 $MCP_CALL "npx -y @anthropic-ai/microsoft-graph-mcp" graph_get_site_drive '{"siteId":"<site-id>"}'

When to Use

• Store audit reports — Upload health check, security audit, and reconciliation reports to SharePoint
• Archive configuration backups — Push running-config snapshots to a versioned SharePoint library
• Retrieve reference documents — Pull network design docs, standards, or runbooks from SharePoint
• Organize artifacts — Create folder structures per site, per device, or per change request
• Search documentation — Find existing topology diagrams, IP address plans, or change records

Integration with Other Skills

• Use pyats-health-check or pyats-security to generate reports, then upload to SharePoint via this skill
• Use netbox-reconcile to produce drift reports, then store them in a dated folder on SharePoint
• Use msgraph-visio to generate Visio diagrams, then organize them in SharePoint document libraries
• Use gait-session-tracking to produce audit logs, then archive them to SharePoint for compliance

Folder Organization Pattern

SharePoint: Network Engineering/
  ├── Health Reports/
  │   ├── 2026-02-22-fleet-health.md
  │   └── 2026-02-21-r1-health.md
  ├── Security Audits/
  │   └── 2026-02-22-cis-audit.md
  ├── Topology/
  │   ├── campus-topology.vsdx
  │   └── dc-fabric.vsdx
  ├── Configuration Backups/
  │   ├── R1/
  │   └── SW1/
  └── Change Records/
      └── CHG0012345-loopback99.md

GAIT Audit Trail

Record file operations in GAIT:

python3 $MCP_CALL "python3 -u $GAIT_MCP_SCRIPT" gait_record_turn '{"input":{"role":"assistant","content":"Uploaded fleet health report (2026-02-22-fleet-health.md) to SharePoint Network Engineering/Health Reports/. 4 devices assessed, 1 WARNING (SW1 CPU 82%), 3 HEALTHY.","artifacts":[]}}'