Silentfail
About
Diagnose MCP servers — health checks, tool testing, token cost audits, conflict detection, and security scanning with 50+ prompt injection patterns. Works as CLI or MCP server inside Claude Desktop.
Explore
- Discovers MCP configs from all major clients.
- Spawns servers and measures startup time.
- Tests each tool with smart inferred parameters.
- Calculates exact token cost per server and tool.
- Detects tool name conflicts across servers.
- Scans for security threats like prompt injection.
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
SilentfailCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
git clone https://github.com/decksaga/silentfail.git
cd silentfail
npm install
npm run build
SilentFail works in two modes: MCP Server (Claude scans itself) or CLI (you run it from the terminal).
<br />
---
<br />
scan_mcp_servers
Full diagnostic: health, tokens, conflicts, security, recommendations. Set `test_tools: true` to verify each tool works.
open_silentfail_dashboard
Run scan + open interactive web dashboard on `localhost:3777`
Result
Meaning
| Tool | Description |
|:-----|:------------|
| scan_mcp_servers | Full diagnostic: health, tokens, conflicts, security, recommendations. Set test_tools: true to verify each tool works. |
| open_silentfail_dashboard | Run scan + open interactive web dashboard on localhost:3777 |
<br />
---
<br />
SilentFail doesn't just check if servers respond — it calls each tool.
It reads the schema, infers valid test inputs (AAPL for stocks, USD/EUR for forex, bitcoin for crypto), calls the tool, and categorizes the result:
| Result | Meaning |
|:-------|:--------|
| Passed | Tool works, returned data |
| Broken | Runtime error — tool is dead |
| Rejected | Tool works but rejected the test input (validation is working correctly) |
| Skipped | Couldn't infer safe params — manual test recommended |
| Timeout | Took too long to respond |
If it says broken, it's broken. No false positives.
<br />
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"silentfail": {
"silentfail": {
"command": "npx",
"args": [
"-y",
"silentfail@latest",
"--mcp"
]
}
}
}
}McpServers
{
"silentfail": {
"command": "npx",
"args": [
"-y",
"silentfail@latest",
"--mcp"
]
}
}What SilentFail does
| | |
|:--|:--|
| Discover | Finds all MCP configs across Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, and project-level files |
| Connect | Spawns each server, measures startup time, lists all tools |
| Test | Calls each tool with smart inferred parameters to verify it actually works |
| Audit | Calculates exact token cost per server and per tool |
| Detect | Finds tool name conflicts across servers |
| Scan | 50+ security patterns — prompt injection, data exfiltration, dangerous commands, encoded payloads |
| Report | Clear, actionable output in CLI, chat, or visual dashboard |
<br />
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
