WAL MCP Server

MCP (Model Context Protocol) server for RTL waveform analysis using WAL (Waveform Analysis Language).

Supported Waveform Formats

• VCD (Value Change Dump)
• FST (Fast Signal Trace)
• Any other formats supported by WAL

Installation and Usage

pip install wal-mcp

Add to your MCP client configuration:

{
  "mcpServers": {
    "wal-mcp": {
      "type": "stdio",
      "command": "wal-mcp-server",
      "args": []
    }
  }
}

For WAL expression syntax and advanced examples, see the WAL documentation.

This package also comes with script to install a skill and sufficient reference material to teach an AI agent how to write WAL. This allows you to use natural language to analyze waveforms.

To install the skill into .claude/skills/ your current directory run:

wal-mcp-install-skills

Or pass a filepath to the script to install elsewhere.

Alternatively, if you are using the Github CLI application you can install the skill using that.

gh install skill ktbarrett/wal-mcp

Development

To set up a development environment, install the dev group and the current package. This is easiest done with uv.

uv sync

There are also pre-commit hooks which will run lints whenever you make a commit. prek is the recommended pre-commit hook runner.

First install prek if you already haven't.

uv tool install prek

Then install the git hooks.

prek install

Testing

pytest is used for testing. Run the following to run all the tests and collect coverage.

pytest

You can better visualize the coverage by using a coverage report format such as html.

coverage html

Credits

Built on WAL (Waveform Analysis Language), a domain-specific language for hardware waveform analysis. See the WAL website for more information.

License

BSD 3-Clause License. See LICENSE for details.