MCP Service for Chepy Library Functions
This project exposes the powerful Chepy data transformation library as an MCP (Model Context Protocol) server. It allows you to access Chepy's tools (like encoding, decoding, and data manipulation) via a single flexible API: the bake
pipeline tool, inspired by CyberChef.
- Single pipeline tool (
bake
): Chain one or more Chepy operations, each with parameters, in a single request - Chepy recipe JSON format: Pipelines are described using Chepy's own recipe serialization, making them compatible with Chepy's import/export and CLI tools
- Resource endpoint: Discover all available Chepy operations and their signatures
- Unittest-based test suite for robust validation
-
Install uv if it is not installed yet.
$ curl -LsSf https://astral.sh/uv/install.sh | sh
-
Clone the repository:
git clone https://github.com/nebucaz/chepy-mcp.git cd chepy-mcp
-
Install dependencies:
uv sync
Run the server with:
$ uv run src/server.py
The bake
tool expects a pipeline in the Chepy recipe JSON format:
{
"input": "hello world",
"recipe": [
{"function": "to_base64", "args": {}},
{"function": "from_base64", "args": {}}
]
}
- Each step in the
recipe
list is an object with afunction
(the Chepy operation name) andargs
(a dictionary of arguments for that function). - This format is fully compatible with Chepy's own recipe import/export and CLI tools.
The response will indicate if the output is text or binary:
{
"type": "text",
"data": "hello world"
}
Fetch the resource endpoint to get all available Chepy operations, their parameter signatures, and descriptions:
- Resource URI:
resource://chepy_operations
Example response:
{
"to_base64": {
"signature": "(alphabet: str = 'standard')",
"description": "Encode the input string to base64"
},
"from_base64": {
"signature": "(alphabet: str = 'standard', remove_non_alpha: bool = True)",
"description": "Decode base64 encoded string"
}
}
- All Chepy operations are available through the pipeline tool; no need to add individual wrappers.
- Add more tests in the
tests/
directory as needed.
- The
bake
tool will automatically detect if the output is text or binary and encode binary as base64. - For a list of valid operations and their parameters, see the
resource://chepy_operations
resource. - Only the
bake
tool and the Chepy operations resource are exposed for maximum flexibility and maintainability. - The pipeline format is fully compatible with Chepy's own recipe serialization and CLI tools.
MIT