MCP Catalogs
Homeopentrace screenshot

opentrace

by adham90·15·Score 42

OpenTrace is an MCP-native observability engine that connects AI coding agents directly to production data via MCP tools.

developer-toolsmonitoringops-infra
2
Forks
1
Open issues
1 mo ago
Last commit
2d ago
Indexed

Overview

OpenTrace provides a self-hosted observability solution specifically designed for AI coding assistants. It features a custom columnar log store optimized for production environments, running on a $4/month VM. The system captures detailed application metrics, errors, and traces, exposing them to AI agents through MCP tools without requiring traditional dashboards. This enables AI assistants to directly investigate production issues, analyze performance bottlenecks, and understand system behavior without context switching.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Debug production errors by having AI agents search logs and analyze error groups
you:Investigate performance bottlenecks through MCP tools for database queries and endpoint metrics
you:Generate tests and safety assessments based on actual production data
you:How does OpenTrace integrate with my existing application?
you:What kind of data can OpenTrace capture from my application?

When to choose this

Choose OpenTrace if you're already using Postgres and want a low-cost, AI-native observability solution that directly connects your coding assistant to production data without UI dashboards.

When NOT to choose this

Don't choose OpenTrace if you need complex dashboards for human analysts, require write access to databases, or need multi-tenant capabilities with isolated data.

Tools this server exposes

12 tools extracted from the README
  • logs

    Full-text log search, distributed trace assembly, N+1 detection

  • errors

    Error grouping by fingerprint, user impact scoring, stack traces

  • database

    Postgres introspection, EXPLAIN plans, lock and connection analysis

  • watches

    Threshold alerts on error rate, latency, request volume

  • overview

    System health, alerts, incident timeline, settings

  • analytics

    Traffic patterns, endpoint performance, time-series analysis

  • code

    Code risk scores, test generation, blast radius

  • healthchecks

    HTTP endpoint monitoring with uptime tracking

  • servers

    Server and process metrics (CPU, memory, GC)

  • connectors

    Manage database connectors (Postgres, MySQL, etc.)

  • setup

    SDK setup assistant — detects framework, provides config

  • deep_capture

    Per-request deep capture: SQL, HTTP, emails, audit trail

Comparable tools

datadog-mcpnewrelic-mcplogtail-mcpprometheus-mcp

Installation

Installation Options

VPS (Hetzner, DigitalOcean, any Linux server)

ssh root@your-server
curl -fsSL https://raw.githubusercontent.com/adham90/opentrace/main/scripts/install.sh | bash

Docker

docker run -d --name opentrace \
  -p 8080:8080 \
  -v opentrace-data:/data \
  -e OPENTRACE_LISTEN_ADDR=0.0.0.0:8080 \
  ghcr.io/adham90/opentrace:latest

Connect to Claude Desktop

Create a .mcp.json file in your project:

{
  "mcpServers": {
    "opentrace": {
      "command": "npx",
      "args": ["@opentrace-sdk/mcp"],
      "env": {
        "OPENTRACE_URL": "https://your-server.com",
        "OPENTRACE_API_KEY": "your-api-key"
      }
    }
  }
}

FAQ

How does OpenTrace integrate with my existing application?
OpenTrace provides SDKs for Ruby and Node.js that capture structured logs, request lifecycle data, error traces, and runtime metrics - all sent as flat JSON via async I/O without blocking your application.
What kind of data can OpenTrace capture from my application?
The SDK captures detailed request lifecycle data including SQL queries with EXPLAIN plans, external API calls, cache hits/misses, view rendering times, email deliveries, file operations, audit trails, and a waterfall timeline - all in one log entry.

On Hacker News

Recent discussion from the developer community.

Compare opentrace with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.