<!-- Generated by SpecStory -->
# Fixing MCP call initialization errors (2025-06-12 18:30:38)
_**User**_
With MCP call
```
{
"method": "tools/call",
"params": {
"name": "capture_screenshot",
"arguments": {
"url": "https://fontlab.com/",
"zoom": 100,
"width": 0,
"height": 0,
"scroll_step": 100,
"scale": 100,
"app": "",
"output_dir": "/Users/adam/Pictures",
"subdirs": false,
"format": "png",
"anim_spf": 0.5,
"html": true,
"max_frames": 0,
"from_selector": ""
},
"_meta": {
"progressToken": 5
}
}
}
```
```
WARNING:root:Failed to validate request: Received request before initialization was complete
MCP error -32602: Invalid request parameters
```
---
_**Assistant**_
Let me help you understand and fix this issue. First, I'll analyze the error message and then help set up the proper initialization for the MCP server.
The error message indicates two problems:
1. "Failed to validate request: Received request before initialization was complete"
2. "MCP error -32602: Invalid request parameters"
This suggests that we need to:
1. Initialize the MCP server properly before making any requests
2. Ensure we're using the correct tool format for screenshot capture
Let me check the codebase to understand how we should properly initialize and use the MCP functionality.
---
<details>
<summary>Searched codebase "MCP server initialization and screenshot capture in mcp.py" • **25** results</summary>
| File | Lines |
|------|-------|
| `src/brosh/cli.py` | L217-231 |
| `src/brosh/mcp.py` | L152-177 |
| `src/brosh/mcp.py` | L87-111 |
| `src/brosh/mcp.py` | L38-87 |
| `src/brosh/mcp.py` | L111-152 |
| `src/brosh/mcp.py` | L1-38 |
| `src/brosh/tool.py` | L105-147 |
| `src/brosh/mcp.py` | L177-182 |
| `src/brosh/tool.py` | L50-105 |
| `src/brosh/tool.py` | L147-192 |
| `src/brosh/tool.py` | L1-50 |
| `src/brosh/mcp.py` | L182-186 |
| `src/brosh/capture.py` | L1-62 |
| `src/brosh/cli.py` | L1-55 |
| `src/brosh/cli.py` | L89-117 |
| `src/brosh/cli.py` | L155-217 |
| `src/brosh/capture.py` | L62-122 |
| `src/brosh/browser.py` | L1-37 |
| `src/brosh/capture.py` | L206-251 |
| `src/brosh/cli.py` | L55-89 |
| `src/brosh/cli.py` | L117-155 |
| `src/brosh/capture.py` | L162-206 |
| `src/brosh/capture.py` | L122-162 |
| `src/brosh/browser.py` | L213-243 |
| `src/brosh/__init__.py` | L1-11 |
</details>
---
Based on the codebase search, I can see that we need to:
1. Initialize the MCP server first
2. Ensure a browser is running in debug mode before making screenshot requests
Let me help you set this up properly. First, we need to start the MCP server:
---
```bash
brosh mcp
```
---
Now that the MCP server is running in the background, we need to ensure a browser is running in debug mode. Let's start Chrome in debug mode:
---
```bash
brosh --app chrome run
```
```
broshadam@Hamlet brosh % brosh --app chrome run
chromium already running on port 9222
broshadam@Hamlet brosh %
```
---
Great! Now both the MCP server and Chrome in debug mode are running. You can now try your screenshot capture request again. The MCP server should be properly initialized and ready to handle the request.
A few notes about your request parameters:
1. The parameters look valid according to the tool's implementation
2. The output directory `/Users/adam/Pictures` is correct as it matches the default user pictures directory
3. The format "png" is supported
4. The scroll_step of 100 is within the valid range (10-200)
5. The scale of 100 is within the valid range (10-200)
Try making your screenshot capture request again now that the server is properly initialized. If you encounter any other issues, please let me know and I'll help you troubleshoot them.
Also, for future reference, you can combine these steps into a single command by using the `brosh` CLI tool directly:
```bash
brosh --app chrome --width 1920 --height 1080 shot "https://fontlab.com/" --scroll_step 100 --format png
```
This would accomplish the same thing without needing to use the MCP protocol directly. But since you specifically want to use the MCP interface, we've set that up for you.
---
