zplCore.comzplCore.com
DEEN
Get a trial

AI & automationMCP

Your AI agent designs, checks and fixes labels on its own

With the MCP server zplcore-mcp, AI agents such as Claude Code, Claude Desktop, GitHub Copilot in VS Code and Cursor work directly with the zplCore engine: write ZPL, render it, read the report, fix it, until the label is right. Offline, on your machine, through the Model Context Protocol.

Agent ↔ zplcore-mcp

Create a 100 × 150 mm GS1 shipping label for this shipment and check it against the GS1 logistics rules.

  1. passedlint0 diagnostics
  2. errors foundcomplygs1-logistic2 errors · 11 notesZPL3104 SSCC: X-dimension 0.375 mm < 0.495 mm · ZPL3105 bar height 13 mm < 31.75 mm
  3. Agent: SSCC barcode set to ^BY4 and 270 dots high, text 3 mm and up, layout adjusted.
  4. passedcomplygs1-logistic0 errors · 0 warnings · 0 notes
  5. passedverify3 of 3 barcodes read · grade A
  6. passedrenderPNG · 100 × 150 mm · 203 dpi
GS1 shipping label 100 × 150 mm rendered by zplCore with recipient, routing, QR code, tracking number and SSCC barcode
Done.

0 errors · all barcodes grade A

GS1-128 (SSCC), Code 128, QR code

Example run, shortened. Label, findings and grades are real engine results; a pre-check, not a certification.
26tools
stdiono network port
offlineno cloud service

26 tools for agents

Create
  • render
  • fill_template
  • format
  • convert_epl
  • import_label
  • lift_svg
Check & simulate
  • lint
  • diff
  • comply
  • verify
  • printer_profiles
  • printer_check
  • print_estimate
  • cut_plan
Data, GS1 & RFID
  • extract
  • redact
  • gs1_link_build
  • gs1_link_parse
  • gs1_migrate
  • rfid_encode
  • rfid_decode
  • rfid_check
Look up
  • list_symbologies
  • describe_symbology
  • explain_command
  • get_limits

Works with

  • Claude Code
  • Claude Desktop
  • VS Code · GitHub Copilot
  • Cursor

Built for untrusted input

  • stdio only, no network port; every call runs in its own worker process with time and memory limits
  • No access to files, network, shell or environment variables; on macOS also inside the sandbox-exec sandbox
  • Text from the label is marked as untrusted data; hardened with adversarial and fuzz tests
  • Automation without AI: the same checks in CI with zplcore check and the GitHub Action
# Claude Code: add the server to the current project
claude mcp add zplcore -- zplcore-mcp

zplcore-mcp ships with the SDK as a dotnet tool (.NET 10). Installation and the setup for Claude Desktop, VS Code and Cursor are in the docs.

MCP docs →Security model →Label tests in CI →