Merge pull request #137 from sitamgithub-MSIT/kitops-mcp

Changes Suggested in KitOps MCP

Author: patchy631

kitops-mcp/README.md

@@ -1 +1,72 @@
-# KitOps MCP Server
+# KitOps MCP Server
+
+We are going to implement an MCP server to orchestrate KitOps for managing and distributing machine learning models. Agents will be able to connect to discover tools for creating, inspecting, pushing, pulling, and removing ModelKits from remote registries like Jozu Hub.
+
+We use:
+
+- [KitOps](https://kitops.org/) for versioning, packaging, and distributing ML models
+- Cursor (MCP Host)
+
+## Set Up
+
+Follow these steps one by one:
+
+### Install Kit CLI
+
+Here is the documentation for downloading and installing the Kit CLI: [Kit CLI Installation](https://kitops.org/docs/cli/installation/) for your operating system.
+
+After installing the Kit CLI, you can verify the installation by running:
+
+```bash
+kit version
+```
+
+### Create .env File
+
+Create a `.env` file in the root directory of your project with the following content:
+
+```env
+JOZU_USERNAME=<your_jozu_hub_email>
+JOZU_PASSWORD=<your_jozu_hub_account_password>
+JOZU_NAMESPACE=<name_of_repository_in_jozu_hub>
+```
+
+All the values are associated with your Jozu Hub account. If you don't have a Jozu account, you can create one at [Jozu Hub](https://jozu.ml/).
+
+### Install Dependencies
+
+```bash
+uv sync
+```
+
+## Use MCP Server
+
+Run the MCP server with the created configuration file as `mcp.json` either globally or in the current project directory. Here's the code of configuring MCP globally to run the server:
+
+```json
+{
+  "mcpServers": {
+    "kitops_mcp": {
+      "command": "uv",
+      "args": [
+        "--directory",
+        "/Users/akshay/Eigen/ai-engineering-hub/kitops-mcp",
+        "run",
+        "--with",
+        "mcp",
+        "server.py"
+      ]
+    }
+  }
+}
+```
+
+## 📬 Stay Updated with Our Newsletter!
+
+**Get a FREE Data Science eBook** 📖 with 150+ essential lessons in Data Science when you subscribe to our newsletter! Stay in the loop with the latest tutorials, insights, and exclusive resources. [Subscribe now!](https://join.dailydoseofds.com)
+
+[![Daily Dose of Data Science Newsletter](https://github.com/patchy631/ai-engineering/blob/main/resources/join_ddods.png)](https://join.dailydoseofds.com)
+
+## Contribution
+
+Contributions are welcome! Feel free to fork this repository and submit pull requests with your improvements.

kitops-mcp/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "kitops-mcp"
 version = "0.1.0"
-description = "Add your description here"
+description = "MCP server for managing KitOps ModelKits remotely"
 readme = "README.md"
 requires-python = ">=3.12"
 dependencies = [

kitops-mcp/server.py

@@ -108,7 +108,7 @@ def remove_modelkit(modelkit_tag: str) -> str:
     This operation cannot be undone, so use it with caution.
 
     Args:
-        modelkit_tag: The tag of the ModelKit to remove in the format 'owner/name:tag'.
+        modelkit_tag: The tag of the ModelKit to remove in the format 'registry/namespace/repository:tag'.
 
     Returns:
         A message confirming the ModelKit was removed successfully.

kitops-mcp/tools.py

@@ -80,13 +80,15 @@ def push_and_pack(modelkit_tag: str, working_dir: str) -> str:
     validate_modelkit_tag(modelkit_tag)
 
     try:
-        # Create a ModelKitManager instance and pack and push the ModelKit
-        kitfile = Kitfile(path=os.path.join(working_dir, "Kitfile"))
-        if not os.path.exists(os.path.join(working_dir, "Kitfile")):
+        # Check if Kitfile exists before creating instance
+        kitfile_path = os.path.join(working_dir, "Kitfile")
+        if not os.path.exists(kitfile_path):
             raise FileNotFoundError(
                 f"Kitfile not found in the working directory '{working_dir}'"
             )
+        kitfile = Kitfile(path=kitfile_path)
 
+        # Create a ModelKitManager instance and pack and push the ModelKit
         manager = ModelKitManager(
             working_directory=working_dir, modelkit_tag=modelkit_tag
         )

kitops-mcp/util.py

@@ -11,7 +11,7 @@ def validate_modelkit_tag(modelkit_tag: str) -> bool:
         )
 
     # Split the tag to check individual components
-    path, tag_part = modelkit_tag.split(":")
+    path, tag_part = modelkit_tag.rsplit(":", 1)
     path_parts = path.split("/")
 
     # Ensure all parts are non-empty