Agentic AI & MCP Server (@stacklenzz/mcp)
Connect your live backend telemetry, active 5xx crash logs, distributed traces, and SLO compliance metrics directly to AI coding assistants like Cursor, Windsurf, Claude Desktop, and Antigravity.
@stacklenzz/mcp, your AI assistant can query live backend errors, inspect function stack traces, check database latency, and suggest code fixes directly in your IDE without leaving your editor.Model Context Protocol Overview
The Model Context Protocol (MCP) is an open standard that allows LLM applications to securely access local tools and real-time operational state. The @stacklenzz/mcp package implements a standard MCP stdio server exposing 5 dedicated telemetry tools:
get_service_health
Returns real-time health score, uptime, active error rate, CPU load, and V8 heap utilization.
get_recent_errors
Retrieves recent 4xx/5xx crash logs, fingerprints, stack traces, and request breadcrumbs.
get_recent_traces
Inspects distributed trace waterfall spans (controller, middleware, database) to pinpoint latency bottlenecks.
get_slo_status
Queries SRE Target compliance (e.g. 99.5%), remaining error budget percentage, and burn rate.
🚀 Quick Start: Running the MCP Server
You can launch the MCP server over standard I/O (stdio) using npx:
npx -y @stacklenzz/mcp
By default, it queries your local backend stats endpoint at http://localhost:5000/api/observability/stats. You can customize the target backend using the STACKLENZZ_ENDPOINT environment variable.
⚙️ IDE & Assistant Configurations
Cursor IDE Setup (.cursor/mcp.json)
Add the following configuration to .cursor/mcp.json in your project workspace:
{
"mcpServers": {
"stacklenzz": {
"command": "npx",
"args": ["-y", "@stacklenzz/mcp"],
"env": {
"STACKLENZZ_ENDPOINT": "http://localhost:5000/api/observability/stats"
}
}
}
}Claude Desktop Setup (claude_desktop_config.json)
Add Stacklenzz to your global Claude Desktop configuration file:
{
"mcpServers": {
"stacklenzz-telemetry": {
"command": "npx",
"args": ["-y", "@stacklenzz/mcp"],
"env": {
"STACKLENZZ_ENDPOINT": "http://localhost:5000/api/observability/stats"
}
}
}
}💬 Example AI Assistant Prompts
Once configured, you can ask your AI coding assistant questions like: