Debug Adapter Protocol (DAP)

by kashuncheng

21 stars
3.7k downloads
Not rated
GitHub

About

Bridges debugging tools with AI systems through the Debug Adapter Protocol, enabling debugger control, breakpoint management, and program state inspection with XML-formatted output for enhanced debugging workflows.

Details

Author
kashuncheng
Repository
KashunCheng/dap_mcp
GitHub stars
21
Downloads
3,711
License
GNU Affero General Public License v3.0
Categories
Productivity, Developer Tools, AI, Knowledge Base, Infrastructure, Other
Tags
#integration

- Debug Adapter Protocol Integration: Interact with debuggers using a standardized protocol.
- MCP Framework: Leverage MCP to optimize context and enhance debugging workflows.
- Rich Debugging Tools: Set, list, and remove breakpoints; control execution (continue, step in/out/next); evaluate expressions; change stack frames; and view source code.
- Flexible Configuration: Customize debugger settings, source directories, and other parameters via a JSON configuration file.

Setting up with Highlight

This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:

  1. Download and install Highlight from highlightai.com/download
  2. Navigate to the plugins tab and select "Add Custom Plugin"
  3. Configure the plugin with the settings below
    Plugin Name Debug Adapter Protocol (DAP)
    Command (node, npx, python, etc.) npx
    Arguments
    • Argument 1 -y
    • Argument 2 @highlight/mcp-server

    Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.

  4. Enable "Start Automatically" if you want the plugin to start when Highlight launches

From the repository

The project uses a JSON configuration file (e.g., .config.json) to specify debugger settings and source directories. An example configuration:

{
  "type": "debugpy",
  "debuggerPath": "/path/to/python/with/debugpy",
  "debuggerArgs": [
    "-m",
    "debugpy.adapter"
  ],
  // source directories for resolving file paths
  // if you always use absolute paths, you can omit this
  "sourceDirs": [
    "/path/to/source/code"
  ],
  // debugger-specific settings start here
  // configurations for debugpy can be found at
  // https://github.com/microsoft/debugpy/wiki/Debug-configuration-settings
  
  // you can use "program" instead of "module" to specify the program to debug
  "module": "pytest",
  // the python executable to use to run the debuggee
  "python": ["/path/to/python"],
  "cwd": "/path/to/working/directory"
}

This configuration informs the debugger about:
- The path to the debugger executable and its arguments.
- The source directories for resolving file paths during breakpoint operations.
- Other settings (such as module, working directory, and interpreter path) necessary for launching the debuggee.

Install dap-mcp and its dependencies:

pip install dap-mcp
python -m dap_mcp --config config.json

uvx dap-mcp@latest --config config.json

launch

Launch the debuggee program.

set_breakpoint

Set a breakpoint at a specified file and line (with an optional condition).

remove_breakpoint

Remove a breakpoint from a specified file and line.

list_all_breakpoints

List all breakpoints currently set in the debugger.

continue_execution

Continue program execution after hitting a breakpoint.

step_in

Step into a function call.

step_out

Step out of the current function.

next

Step over to the next line of code.

evaluate

Evaluate an expression in the current debugging context.

change_frame

Switch to a different stack frame.

view_file_around_line

View source code around a specified line (using the last provided file if none is specified).

terminate

Terminate the debugging session.

The project exposes several tools that can be invoked via the MCP framework:

- launch: Launch the debuggee program.
- set_breakpoint: Set a breakpoint at a specified file and line (with an optional condition).
- remove_breakpoint: Remove a breakpoint from a specified file and line.
- list_all_breakpoints: List all breakpoints currently set in the debugger.
- continue_execution: Continue program execution after hitting a breakpoint.
- step_in: Step into a function call.
- step_out: Step out of the current function.
- next: Step over to the next line of code.
- evaluate: Evaluate an expression in the current debugging context.
- change_frame: Switch to a different stack frame.
- view_file_around_line: View source code around a specified line (using the last provided file if none is specified).
- terminate: Terminate the debugging session.

These tools provide XML-rendered output for integration with MCP clients.

Claude Desktop / Cursor

Paste into your MCP client config file to install this server.

{
    "mcpServers": {
        "debug adapter protocol (dap)": {
            "env": {},
            "args": [
                "-y",
                "@highlight/mcp-server"
            ],
            "command": "npx"
        }
    }
}

Linux

{
    "env": [],
    "args": [
        "-y",
        "@highlight/mcp-server"
    ],
    "command": "npx"
}

Macos

{
    "env": [],
    "args": [
        "-y",
        "@highlight/mcp-server"
    ],
    "command": "npx"
}

Windows

{
    "env": [],
    "args": [
        "/c",
        "npx",
        "-y",
        "@highlight/mcp-server"
    ],
    "command": "cmd"
}

dap-mcp

dap-mcp is an implementation of the Model Context Protocol (MCP) tailored for managing Debug Adapter Protocol (DAP) sessions. MCP provides a standardized framework to optimize and extend the context window of large language models, and in this project, it is used to enhance and streamline debugging workflows.

Features

- Debug Adapter Protocol Integration: Interact with debuggers using a standardized protocol.
- MCP Framework: Leverage MCP to optimize context and enhance debugging workflows.
- Rich Debugging Tools: Set, list, and remove breakpoints; control execution (continue, step in/out/next); evaluate expressions; change stack frames; and view source code.
- Flexible Configuration: Customize debugger settings, source directories, and other parameters via a JSON configuration file.

Installation

Prerequisites

- Python 3.10 or higher
- uv (optional, for running the server)

Installing and Running the Server

Install dap-mcp and its dependencies:

pip install dap-mcp
python -m dap_mcp --config config.json

Or, if you have uv installed

uvx dap-mcp@latest --config config.json

Configuration

The project uses a JSON configuration file (e.g., .config.json) to specify debugger settings and source directories. An example configuration:

{
  "type": "debugpy",
  "debuggerPath": "/path/to/python/with/debugpy",
  "debuggerArgs": [
    "-m",
    "debugpy.adapter"
  ],
  // source directories for resolving file paths
  // if you always use absolute paths, you can omit this
  "sourceDirs": [
    "/path/to/source/code"
  ],
  // debugger-specific settings start here
  // configurations for debugpy can be found at
  // https://github.com/microsoft/debugpy/wiki/Debug-configuration-settings
  
  // you can use "program" instead of "module" to specify the program to debug
  "module": "pytest",
  // the python executable to use to run the debuggee
  "python": ["/path/to/python"],
  "cwd": "/path/to/working/directory"
}

This configuration informs the debugger about:
- The path to the debugger executable and its arguments.
- The source directories for resolving file paths during breakpoint operations.
- Other settings (such as module, working directory, and interpreter path) necessary for launching the debuggee.

Available Debugger Types

| Type | Example Path | Example Args | |---------|---------------------|-----------------------------| | debugpy | /usr/bin/python3 | ["-m", "debugpy.adapter"] | | lldb | /usr/bin/lldb-dap | [] |

Available Tools

The project exposes several tools that can be invoked via the MCP framework:

- launch: Launch the debuggee program.
- set_breakpoint: Set a breakpoint at a specified file and line (with an optional condition).
- remove_breakpoint: Remove a breakpoint from a specified file and line.
- list_all_breakpoints: List all breakpoints currently set in the debugger.
- continue_execution: Continue program execution after hitting a breakpoint.
- step_in: Step into a function call.
- step_out: Step out of the current function.
- next: Step over to the next line of code.
- evaluate: Evaluate an expression in the current debugging context.
- change_frame: Switch to a different stack frame.
- view_file_around_line: View source code around a specified line (using the last provided file if none is specified).
- terminate: Terminate the debugging session.

These tools provide XML-rendered output for integration with MCP clients.

Extending with Other DAP Servers

To support additional DAP servers, you can simply add a new DAP-specific configuration class in the dap_mcp/config.py file. All DAP configurations extend from the base DAPConfig class. Each new subclass should:

- Define a unique type value (using a Literal) to act as a discriminator.
- Include any additional fields or settings specific to that debugger.

For example, to add support for a hypothetical DAP server called "mydap", you might add:

class MyDAP(DAPConfig):
    type: Literal["mydap"]
    # Add any additional settings for MyDAP here
    customSetting: Optional[str] = Field(
        None, description="A custom setting for MyDAP."
    )

After creating your new configuration class, update the union type used for debugger-specific configurations by including your new class. For example:

DebuggerSpecificConfig = Annotated[Union[DebugPy, MyDAP], Field(..., discriminator="type")]

Now, when you supply a configuration JSON with "type": "mydap", it will be parsed and validated using your new MyDAP class, and your DAP server extension will be fully integrated.

Contributing

Contributions are welcome! To contribute:
1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Write tests and ensure all checks pass.
4. Submit a pull request.

Please follow the coding guidelines and include appropriate tests with your changes.

License

This project is licensed under the AGPL-3.0 License. See the LICENSE file for details.

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.