mcpmock
CLI tool that generates a mock Model Context Protocol server from list of cases in a YAML file.
## Installation
From npm: `npm install -g @strowk/mcpmock`
From Github Releases: [Download](https://github.com/strowk/mcpmock/releases), unpack and put binary in your PATH
From sources:
```bash
go get github.com/strowk/mcpmock
go install github.com/strowk/mcpmock
```
## Usage
For example if you define something like this in a YAML file:
```yaml
case: List tools
# requesting list of tools
in: {"jsonrpc": "2.0", "method": "tools/list", "id": 1}
# expect one tool in the list
out: {"jsonrpc": "2.0", "result":{ "tools": [{"description": "Hello MCP", "inputSchema": {"type": "object"}, "name": "hello"}] }, "id": 1}
---
case: Call Hello tool
# calling the tool
in: {"jsonrpc": "2.0", "method": "tools/call", "params": {"name": "hello", "arguments": {}}, "id": 1}
# expect "Hi!" as output
out: {
"jsonrpc": "2.0",
"id": 1,
"result": {
"content": [
{"type": "text", "text": "Hi!"}
],
"isError": false
},
}
```
Then if you put it in the folder `testdata` (make sure file also ends with `_test.yaml`) and run mcpmock like this:
```bash
mcpmock serve testdata
```
It would start a mock MCP server with stdio transport that would serve the cases defined in the YAML file.
If you now copy and paste this into your terminal:
```json
{"jsonrpc": "2.0", "method": "tools/list", "id": 1}
```
, you should see list of tools as is defined in first MCP case.
And sending this:
```json
{"jsonrpc": "2.0", "method": "tools/call", "params": {"name": "hello", "arguments": {}}, "id": 1}
```
, you should get the output as defined in second case.
It would also take care to replace the `id` field in the response with the `id` from the request, so if you send this:
```json
{"jsonrpc": "2.0", "method": "tools/call", "params": {"name": "hello", "arguments": {}}, "id": 2}
```
, you should get the output as defined in second case, but with `id` field set to `2`.