Skip to main content
Glama

MCP for Vivado

An AI-oriented Vivado MCP workbench.
This project adopts an MCP-first design and does not provide a self-developed desktop GUI; Vivado-related operations must be performed in the actual Vivado GUI, switching to the corresponding visual context at critical steps.

Current Status

  • Version: v2.0.0

  • PM Conclusion: v2 overall completion

  • Current release scope:

    • Windows mainline usage scenarios

    • v2 public workflow methodology

    • phase-2 core cross-domain closed-loop

    • phase-3 proof / contract / semantic / referent / public schema closing chain

Related MCP server: Vivado MCP

Applicable Scenarios

  • AI-driven MCP / workflow methods to manage the main Vivado design, verification, and debugging processes

  • Automatic discovery of local Vivado versions and workspace-level binding

  • Designing IP, designing Block Design, and running simulations

  • Automatic feedback and re-verification across simulation / timing / ILA / design follow-up

  • Displaying context corresponding to key steps in the Vivado GUI

Directory Description

Quick Start

  1. Install Python 3.11+

  2. Install Vivado and ensure the version is no lower than 2018.3

  3. Execute:

cd D:\ai_playgraund\vivado_mcp
powershell -ExecutionPolicy Bypass -File .\scripts\install_mcp_for_ai.ps1
  1. Follow docs/USER_INSTALL_AND_USAGE.md or docs/AI_INSTALL_AND_USAGE.md to start using

Current Entry Points

The current release exposes Python in-process MCP-style entry points:

Core classes:

  • fpga_mcp.FpgaMcpServer

  • fpga_mcp.FpgaMcpV2PublicServer

Notes:

  • This repository provides a Python package entry point for "method registration + protocol return shell + orchestration wiring"

  • It is not an independent stdio / HTTP transport layer service

  • To connect to an external AI host, you can directly import FpgaMcpV2PublicServer in the host and call list_methods() / invoke()

v2 Public Methods

  • create_workflow

  • get_workflow

  • list_workflows

  • design_block_design

  • design_ip

  • run_simulation

  • confirm_workflow

  • reject_workflow

  • cancel_workflow

  • submit_workflow_input

  • get_workflow_events

Available Tools

15 tools
cancel_workflowC

Cancel a workflow.

Inputs: params. Returns: object

confirm_workflowD

Confirm a workflow.

Inputs: params. Returns: object

create_workflowD

Create a workflow record.

Inputs: params. Returns: object

design_block_designC

Design a block design workflow intent.

Inputs: params. Returns: object

design_ipC

Design an IP workflow intent.

Inputs: params. Returns: object

get_workflowC

Get a workflow snapshot.

Inputs: params. Returns: object

get_workflow_eventsC

Get public workflow events.

Inputs: params. Returns: object

list_public_methodsB

List released v2 public FPGA MCP methods.

Inputs: no named parameters. Returns: result

list_registered_toolchainsA

List registered Vivado toolchains available to stdio workflows.

Inputs: no named parameters. Returns: result

list_workflowsD

List workflows.

Inputs: params. Returns: object

register_toolchain_pathB

Register another Vivado install path or vivado.bat path and persist it for future sessions.

Inputs: params. Returns: object

reject_workflowC

Reject a workflow.

Inputs: params. Returns: object

remove_registered_toolchainB

Remove a previously registered Vivado toolchain option from stdio workflows.

Inputs: params. Returns: object

run_simulationD

Run a simulation workflow intent.

Inputs: params. Returns: object

submit_workflow_inputC

Submit workflow user input.

Inputs: params. Returns: object

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.8/5.0

    Scored across 15 tools

    Disambiguation4/5

    Most tools have distinct purposes (e.g., workflow CRUD, toolchain management, design intents), but 'cancel_workflow' and 'reject_workflow' could be ambiguous—both involve stopping a workflow but with different semantics. Descriptions are brief but generally clear.

    Naming Consistency4/5

    Tool names follow a consistent verb_noun pattern (e.g., create_workflow, list_workflows). Minor inconsistency: 'design_block_design' is redundant (block design appears twice), and 'run_simulation' uses a different verb style than 'design_*' intents.

    Tool Count5/5

    15 tools is well-scoped for a Vivado MCP server covering workflow lifecycle, design intents, toolchain registration, and public methods. Each tool serves a clear purpose without unnecessary bloat.

    Completeness4/5

    Covers CRUD for workflows, toolchain registration, and design intents. Minor gaps: lacks update/delete for registered toolchains (only remove), and no tool for listing or managing IP/block designs after creation. Core workflows are complete.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      B
      quality
      D
      maintenance
      A Model Context Protocol server that lets AI clients operate AMD Vivado through safe, workflow-level tools while allowing the user to watch and interact with the Vivado GUI.
      21
      1
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      A lean MCP server that lets AI agents drive Xilinx Vivado for synthesis, implementation, simulation, and diagnostics through 30 curated tools and raw Tcl execution.
      Apache 2.0
    • F
      license
      Not graded
      quality
      B
      maintenance
      MCP server for AMD Xilinx Vivado 2023.1 that enables Claude Code to act as an AI FPGA engineer, providing 27 tools for querying, building, simulating, and programming FPGA designs.
      2
      -