Skip to content

Conversation

@fgaray
Copy link
Contributor

@fgaray fgaray commented Nov 21, 2025

Summary

This PR introduces a Model Context Protocol (MCP) server integration for OpenROAD. This allows OpenROAD functionality to be exposed to and controlled by MCP-compliant clients (such as LLM agents). It establishes
the basic server infrastructure, exports core OpenROAD commands, and adds utilities for output capturing and database management.

Key Changes

  • Infrastructure Setup:
    • Added fastmcp dependency.
    • Created the basic MCP server implementation in openroad/mcp.
  • Functionality:
    • Exported basic OpenROAD commands to the MCP interface.
    • Added capability to capture stdout into strings for easier processing.
    • Implemented a command to reset the OpenROAD database.
    • Clarified the execution environment context for the agent.
  • Documentation:
    • Added a README for the MCP server component to explain setup and usage.

Commits

  • Add readme of the mcp server
  • Making more explicit to the agent where OpenROAD is running
  • Allowing capturing stdout into strings and reseting the OpenROAD's database
  • Basic export of OpenROAD comamnds to MCP
  • mcp: Add fastmcp server infrastructure

Demo

https://gist.github.com/fgaray/5e944d829201a41fc0610073d822407f

Herre is a demo of gemini-cli controlling OpenROAD and permforming a timing analysis of the design. It suggest actions to take, explores the design and can interpret the results.

Added fastmcp dependency and created a basic MCP server implementation
in openroad/mcp. This sets up the infrastructure for exposing
OpenROAD functionality via the Model Context Protocol.

Signed-off-by: Felipe Garay <[email protected]>
Signed-off-by: Felipe Garay <[email protected]>
@github-actions
Copy link
Contributor

clang-tidy review says "All clean, LGTM! 👍"

Signed-off-by: Felipe Garay <[email protected]>
@github-actions
Copy link
Contributor

clang-tidy review says "All clean, LGTM! 👍"

Signed-off-by: Felipe Garay <[email protected]>
@github-actions
Copy link
Contributor

clang-tidy review says "All clean, LGTM! 👍"

@maliberty maliberty requested a review from luarss November 21, 2025 01:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant