GitHub - gornskew/lisply-mcp: Model Context Protocol (MCP) server to manage and talk to compliant "Lisply" lisp-speaking backend services

Note: This wrapper does not start or manage containers. It is a pure HTTP client to an already-running Lisply backend. Container lifecycle is owned by docker compose — see the "Containerized Runnings" section of the skewed-emacs README — or run your own Lisply backend directly on a host and point the wrapper at its host/port. Once a backend is running, Claude Desktop connects to it according to the example configurations below.

Model Context Protocol (MCP) Middleware for Lisp-based and Lisp-like Environments

This project is a Model Context Protocol (MCP) middleware that enables Large Language Models (LLMs) to interact with Lisp-based development and runtime environments using a lightweight protocol called Lisply.

Who Is this Meant For?

• AI practitioners curious about Lisp
• Lisp practitioners curious about AI
• Anyone interested in Neuro-Symbolic Programming
• Mechanical/Civil Engineers and Designers interested in CAD Automation and Knowledge Based Engineering
• Tinkerers, meddlers, and tamperers from all walks of life

What Is it Meant to Do?

The Lisply-MCP middleware connects MCP-capable AI Agent programs, or MCP Clients, such as ClaudeDesktop, to Lisp-based systems which support a REPL, or Read-Eval-Print Loop. The connection is meant to facilitate AI-assisted symbolic programming sometimes referred to as Neuro-Symbolic Programming. We have coined the term "Lisply" to refer to a lightweight protocol which most any Lisp-like system can implement to render it compatible with this Lisply-MCP middleware.

The idea is that the LLM will be able to generate and evaluate arbitrary Lisp expressions, including creating, compiling, loading, and testing entire files and projects.

Sandbox Trust Model

Lisply-backed MCP servers are intended to be exposed to the LLM as trusted sandboxes. The wrapper is not designed to restrict Lisp operators, filesystem access, or subprocess execution inside the backend environment. Instead, the backend itself is expected to run in an isolated container or other sandbox chosen by the operator.

This is intentional:

• lisp_eval is meant to support free-form, full use by LLMs.
• /projects may be a host-mounted working tree, but the rest of the backend filesystem may remain container-ephemeral.
• Trust decisions should therefore be made at the container/backend boundary, not by crippling Lisp evaluation in the MCP wrapper.

The wrapper now advertises this trust model in tool metadata with a default TRUST_AS_SANDBOX=true. Operators can override the explanatory text with SANDBOX_NOTE if needed.

Extra Quick Start

Follow the "Containerized Runnings" section of the skewed-emacs README.

This will get you a Docker Compose setup including a preconfigured containerized version of Lisply-MCP.

Quick Start

The following will get you up and running quickly with a minimal default configuration and a default public Common Lisp based backend running as a Docker container. See the main Contents below for more background and detailed configuration options.

1. Install

1. Install Node.js (18+ recommended). If on Windows, this can be installed directly in Windows or in WSL.

2. Have a running Lisply backend to connect to. The easiest way is the docker compose stack from the skewed-emacs README (requires Docker); alternatively, run any Lisply-compliant backend directly on your host.

3. Clone this lisply-mcp repository to a location where your MCP-capable AI Agent (e.g. Claude Desktop) can access it.

2. Configure your MCP-capabile AI Agent

Edit or create your AI Agent's configuration file as shown below. In the case of Claude Desktop, the configuration file is typically:

/mnt/c/Users/<user>/AppData/Roaming/Claude/claude_desktop_config.json

or

c:\Users\<user>\AppData\Roaming\Claude\claude_desktop_config.json

In the example below, replace /path/to/cloned/ with the correct path to the ./scripts/mcp-wrapper.js file from the cloned repo:

{
  "mcpServers": {
    "gendl-ccl": {
      "command": "node", 
      "args": [
        "/path/to/cloned/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "gendl-ccl",
	"--http-port", "9080"
      ]
    }
  },
  {
    "gendl-sbcl": {
      "command": "node", 
      "args": [
        "/path/to/cloned/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "gendl-sbcl",
	"--http-port", "9090"
      ]
    }
  },

  { 
    "skewed-emacs": {
      "command": "node", 
      "args": [
        "/path/to/cloned/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "skewed-emacs",
	"--http-port", "7080"
      ]
    }
  }

  
}

Or in a WSL scenario (where the Claude Desktop is running in the Windows host):

{
  "mcpServers": {
    "gendl-ccl": {
      "command": "wsl", 
      "args": [
        "node /path/to/cloned/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "gendl-ccl",
	"--http-port", "9080"
      ]
    }
  },
  {
    "gendl-sbcl": {
      "command": "wsl", 
      "args": [
        "node /path/to/cloned/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "gendl-sbcl",
	"--http-port", "9090"
      ]
    }
  },

  { 
    "skewed-emacs": {
      "command": "wsl", 
      "args": [
        "node /path/to/cloned/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "skewed-emacs",
	"--http-port", "7080"
      ]
    }
  }
  
}

See the main Contents below for further configuration options, for example how to specify an alternative Lisply backend service host/port. (Sharing/mounting host directories into containerized backends is configured in the docker compose setup, e.g. skewed-emacs compose-dev, not by this wrapper.)

Each server operates independently, allowing you to work with multiple Lisp environments simultaneously without tool name conflicts.

3. Restart your AI Agent and Test

With the above configuration in place, your freshly restarted AI Agent will now have access to an MCP server called lisply-gendl, with a gendl__lisp_eval MCP tool (among a few other tools discussed in the main Contents below). Note that tools are automatically prefixed with the server name to avoid conflicts when running multiple Lisply servers.

In order to test your setup, you can prompt your LLM as follows:

Evaluate (+ 1 2 3) using the gendl__lisp_eval tool, and let me know the result.

The LLM should invoke the requested evaluation and respond with 6 as expected. Feel free to experiment with more complex expressions before proceeding.

How Does the Default Minimal Configuration Work?

The minimal default configuration described in the Quick Start above connects to an already-running Gendl backend (for example from the skewed-emacs docker compose stack), a Common Lisp superset sporting a standard REPL (Read-Eval-Print Loop). The wrapper itself never pulls or starts containers. Note a second Lisply backend implementation for Emacs lisp also exists, within the Skewed Emacs project.

System Overview

The Lisply MCP middleware is implemented as a Javascript program meant to run in Node.js, and provides a bridge between your AI Agent and any compliant Lisply backend system. This wrapper enables the AI Agent to:

1. Evaluate Lisp code in the Lisply Backend and receive the results.
2. Make HTTP requests to any web endpoints implemented in the backend.
3. Access introspection and documentation lookup facilities in the LB using Lisp evaluation.
4. Create, manipulate, compile, load, and analyze files, again using Lisp evaluation.

Lisply is a lightweight protocol that specifies a minimal yet flexible set of HTTP interfaces, a standard set of environment variables, and several optional capabilities to facilitate AI agents controlling your running Lisp system.

Architecture

The diagram below roughly captures how the components interact:

flowchart TB
    User("User") <--> Claude("Claude Desktop")
    User <-.-> Emacs("Emacs Text Editor (Optional)")

    Claude <--> MCP("MCP Protocol")
    MCP <--> Wrapper("Node.js MCP Wrapper")

    Wrapper --> LisplyHttp("Lisply HTTP Server")
    
    subgraph Backend ["Lisply Backend (container or host process)"]
    subgraph LisplyExec["Lisply Executable"]
    LisplyHttp
    LisplySwank("Lisply SWANK Server (for Emacs connection)")
    end
    end
    
    Emacs <-.-> LisplySwank
    
    KB[("Lisply Knowledge Base")] <--> Wrapper
    
    LisplyHttp --> Endpoints("RESTful Endpoints")
    LisplyHttp --> LispEval("Lisp Evaluation")
    
    style User fill:#ff9,stroke:#333,stroke-width:2px
    style Claude fill:#f9f,stroke:#333,stroke-width:2px
    style Emacs fill:#9ff,stroke:#333,stroke-width:2px,stroke-dasharray:5
    style Wrapper fill:#bbf,stroke:#333,stroke-width:2px
    style MCP fill:#bbf,stroke:#333,stroke-width:1px
    style Backend fill:#bfb,stroke:#333,stroke-width:2px
    style LisplyExec fill:#8f8,stroke:#333,stroke-width:2px
    style LisplyHttp fill:#bfb,stroke:#333,stroke-width:1px
    style LisplySwank fill:#bfb,stroke:#333,stroke-width:1px
    style KB fill:#bfb,stroke:#333,stroke-width:1px
    style Endpoints fill:#bfb,stroke:#333,stroke-width:1px
    style LispEval fill:#bfb,stroke:#333,stroke-width:1px

The middleware handles:

1. Translating Lisp evaluation requests between the MCP protocol and the backend Lisply API
2. Error handling and logging

Security Considerations

Because Lisply-MCP allows arbitrary Lisp code to be evaluated against a running Lisp-based backend, there are certain risks in case the LLM were to go "haywire." Therefore, best practices are:

• Allow the wrapper to connect only to a containerized version of a Lisply backend. If overriding default host/port, the wrapper will happily connect to any live Lisply-compliant http port. Avoid allowing this to happen for any http ports being served by programs running directly on your host.

• Make sure not to mount any non-expendable directories into that container (directory mounting is configured in your docker compose setup, not by this wrapper)

• Consider taking steps to limit RAM and CPU usage of the container.

Code Modules/Files

• lib/config.js: Configuration loading and environment handling
• lib/logger.js: Logging functionality
• lib/server.js: HTTP server and MCP wrapper implementation
• lib/utils.js: Utility functions for response handling
• handlers/: Tool-specific request handlers
  • initialize.js: Initialization handler
  • toolsList.js: Tools list handler
  • toolCall.js: Main tool call dispatcher
  • httpRequest.js: HTTP request handler
  • ping.js: Ping handler
  • lispEval.js: Lisp evaluation handler
• mcp-wrapper.js: <--- Main entry point <---

Detailed Installation

1. Clone this repository:

git clone https://github.com/gornskew/lisply-mcp.git

2. Install the required dependencies (optional, as the wrapper auto-installs dependencies):

cd lisply-mcp/scripts
npm install # optional - the script will attempt to do this also if needed
chmod +x mcp-wrapper.js # needed on some systems

3. Test the script:

node mcp-wrapper.js --help

Advanced Configuration

Optional settings for advanced users, with defaults suitable for most cases:

Command-Line Arguments

Options:
  -H, --backend-host <host>            Lisply backend host (default: 127.0.0.1)
  --http-host-port <port>              Backend HTTP port as published on this host; used when
                                       backend host is localhost/loopback (default: 9081)
  --http-port <port>                   Backend HTTP port inside the container network; used for
                                       non-local backend hosts (default: 9080)
  --swank-host-port <port>             SWANK port on host system (documentation/diagnostics) (default: 4201)
  --swank-port <port>                  SWANK port inside container (documentation/diagnostics) (default: 4200)
  --log-file <path>                    Path to log file (default: /tmp/lisply-mcp-wrapper.log)
  --debug                              Enable debug logging
  --endpoint-prefix <prefix>           Prefix for all endpoints (default: lisply)
  --lisp-eval-endpoint <n>             Endpoint name for Lisp evaluation (default: lisp-eval)
  --http-request-endpoint <n>          Endpoint name for HTTP requests (default: http-request)
  --ping-endpoint <n>                  Endpoint name for ping (default: ping-lisp)
  --server-name <name>                 MCP server name for tool prefixing (default: lisply-mcp)
  --eval-timeout <ms>                  Timeout for Lisp evaluation in milliseconds (default: 30000)
  --request-timeout-ms <ms>            Timeout for backend HTTP requests in milliseconds (default: 10000)
  -h, --help                           Display help for command

Environment Variables

The script also supports configuration via environment variables. You can specify variables with the "LISPLY_" prefix or with no prefix:

Note: It is important to keep straight the difference between host ports (listening on and reachable from the host system) and container ports (internal to the container, visible to the Lisply backend service process):

Container Lifecycle: Docker Compose, Not This Wrapper

Earlier versions of this wrapper could pull, start, and manage backend Docker containers itself (image selection, volume mounting, auto-start, existing-service detection). That entire subsystem has been removed. The wrapper is now a pure HTTP client: it connects to whatever Lisply backend is already listening at the configured host and port, and reports a helpful error (with a compose hint) when nothing is there.

For a containerized backend stack (Gendl, skewed-emacs, etc.), use the docker compose setup described in the skewed-emacs README (compose-dev), which owns image selection, volume mounting, port publishing, and UID mapping. For a non-containerized backend, start any Lisply-compliant server yourself (e.g. the host-Emacs path described in skewed-emacs docs/HOST_EMACS_MCP.md) and point the wrapper at its host and port.

Communication

Two links are involved, and they are easy to conflate:

1. AI Agent ↔ wrapper: MCP protocol over standard input/output (the standard MCP stdio transport). This is JSON-RPC plumbing managed by your MCP client, unrelated to any backend REPL.

2. Wrapper ↔ backend: HTTP only. The wrapper POSTs to the backend's Lisply HTTP endpoints and returns structured responses.

Characteristics of the HTTP backend link:

• Structured responses with separate result, stdout, and error fields
• Errors are trapped by the backend and returned as strings
• Response format: {Result: <result>, Stdout: <output>, Error: <any error>}

Example response:

{"Result": "6", "Stdout": "This is a message to standard output"}

An earlier "stdio mode", which talked to a wrapper-started container's raw REPL (interactive debugger, incremental output), was removed along with container auto-starting. Equivalent capabilities may return at the HTTP layer in the future (e.g. a restarts endpoint, streamed output) without re-coupling the wrapper to container lifecycle.

Usage Examples

All the examples below can be tested on command line and used in claude_desktop_config.json configuration (see the Quick Start configuration examples above).

Adding a Separate, Compatible Filesystem MCP Server

Below is a claude_desktop_config.json which sets up a filesystem mcp server as well as our lisply-gendl server. (The filesystem server gets its mount here; any mounts into a containerized Lisply backend are configured in the compose setup, not by this wrapper.)

{
  "mcpServers": {
    "filesystem": {
      "command": "wsl",
      "args": [
        "docker",
        "run",
        "-i",
        "--rm",
        "-u",
        "1000:1000",
        "--mount",
        "type=bind,src=/home/user/projects,dst=/projects",
        "mcp/filesystem",
        "/projects"
      ]
    },
    "lisply-gendl": {
      "command": "wsl",
      "args": [
        "node",
        "/home/user/projects/lisply-mcp/scripts/mcp-wrapper.js",
        "--server-name", "gendl"
      ]
    }
  },
  "globalShortcut": ""
}

Tool Details for Claude

Lisp Evaluation Tool (<server>__lisp_eval)

The lisp_eval tool (prefixed with the server name, e.g., gendl__lisp_eval) allows Claude to evaluate Lisp code directly within the Lisply environment with these parameters:

• code (required): The Lisp code to evaluate
• package (optional): The package to use for the evaluation

HTTP Request Tool (<server>__http_request)

The http_request tool (prefixed with the server name, e.g., gendl__http_request) enables the AI Agent to interact with any HTTP endpoint exposed by the Lisply backend (the following example works with our default backend, which has a built-in /color-map http endpoint):

{
  "path": "/color-map",
  "method": "GET",
  "headers": {
    "Accept": "application/json"
  }
}

Ping Tool (<server>__ping_lisp)

The ping_lisp tool (prefixed with the server name, e.g., gendl__ping_lisp) lets Claude confirm that the Lisply server is running:

Real-World Examples

Example 1: Creating a Parametric Box

In this example, Claude creates a parametric box in the Lisply environment and then modifies its properties based on user requirements.

User: "Can you create a box with width 10, height 5, and depth 3? Then show me how to change the height to 8."

Claude:

I'll create a parametric box using the lisp_eval tool, then show you how to modify its height.

First, let's create the box with the specified dimensions:

(in-package :gdl-user)

;; First, create a box with the specified dimensions
(defparameter *my-box* (make-object 'box 
                         :width 10 
                         :height 5 
                         :length 3))

;; Let's verify the dimensions
(list (theo *my-box* width)
      (theo *my-box* height)
      (theo *my-box* length))

Now, let's change the height from 5 to 8:

;; Change the height using setf with accessor
(theo *my-box* (set-slot! :height 8))

;; Alternatively, we could create a new box
(defparameter *new-box* (make-object 'box 
                         :width 10 
                         :height 8  ;; Updated height
                         :length 3))

;; Verify the new dimensions of original box
(list (theo *my-box* width)
      (theo *my-box* height)
      (theo *my-box* length))

This box can be visualized by creating a web interface or using a built-in gendl visualization tool such as geysr. See Gendl Documentation for details.

Troubleshooting

Common Issues and Solutions

Backend Not Running

If the wrapper reports it cannot reach the backend:

1. For the containerized stack, make sure the compose services are up:

cd ~/projects/skewed-emacs && ./compose-dev up

2. Check whether anything is listening on the expected port:

curl http://localhost:9081/lisply/ping-lisp

Connection Errors

If the LLM Agent / MCP Client cannot connect to the configured Lisply backend:

1. Check if the Lisply server is running (for the compose stack):

cd ~/projects/skewed-emacs && ./compose-dev ps

2. Check the wrapper's log file:

tail -f /tmp/lisply-mcp-wrapper.log

3. Check the Claude Desktop log file with Windows tools e.g. Notepad. This is typically in a location such as:

WSL/Linux:

/mnt/c/Users/<user>/AppData/Roaming/Claude/logs/mcp-server-lisply.log

Windows:

c:\Users\<user>\AppData\Roaming\Claude\logs\mcp-server-lisply.log

5. Try curling to the Lisply HTTP server:

curl http://localhost:9081/lisply/ping-lisp

6. Try connecting to the Lisply SWANK server (on default port 4201):

M-x slime-connect  ;; from emacs

Note that the setting up the Skewed-Emacs configuration will enable M-x slime-connect in your emacs.

Permission Issues

If you encounter file-ownership surprises in a directory mounted into a containerized backend, remember mounts and UID mapping are configured in the compose setup, not by this wrapper. Check the mounted directory permissions:

ls -l /path/to/mounted/directory

Diagnostic Commands

Use these commands to diagnose general issues:

1. Check the middleware logs:

tail -f /tmp/lisply-mcp-wrapper.log

2. Check backend container logs (for the compose stack):

cd ~/projects/skewed-emacs && ./compose-dev logs

3. Check Lisply service status:

curl http://localhost:9081/lisply/ping-lisp

4. Verify Docker environment:

License

This software is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0), the same license used by Gendl.

License Implications

Simply using this MCP server to interact with a Lisply backend and obtain outputs does not trigger the requirements of the AGPL, e.g. you can use this wrapper to interact with Gendl without being required to share your code.

However, if you modify or extend this wrapper, or a license-compatible Lisply backend such as Gendl, and wish to distribute and/or host a service based on that result (commercial or not), then the AGPL would require you to share your modifications with the downstream recipients or users.

For applications that need to keep their source code closed, Genworks has begun offering an "escape clause" from AGPL restrictions for a 5% self-reported quarterly revenue royalty. More information and a payment gateway are available at royalties.genworks.com.

The full text of the license can be found in the COPYING.txt file in this directory.

MCP Server Registries

• MCPHub