I built OpsOrch, an open-source orchestration layer that gives you one unified API for incidents, logs, metrics, tickets, messaging, and service metadata. It sits on top of the tools you already use (PagerDuty, Jira, Elasticsearch, Prometheus, Slack, etc.) and normalizes everything into a single schema.

OpsOrch doesn’t store your operational data. It just brokers requests through pluggable adapters (Go or JSON-RPC) and returns unified structures. On top of this, there’s an optional MCP server that exposes all capabilities as typed tools for LLM agents.

Why? Most incident workflows today require jumping across 5+ vendor UIs and APIs, each with its own query language and auth model. OpsOrch aims to be the small, transparent “glue layer” that removes that complexity without forcing a migration.

What’s available now:

Core orchestration service (Go, Apache-2.0)

Adapters: PagerDuty, Jira, Prometheus, Elasticsearch, Slack, plus mock providers

MCP server exposing incidents/logs/metrics/tickets/services as agent tools

No vendor lock-in, no data gravity

Repos: https://github.com/OpsOrch/opsorch-core

https://github.com/OpsOrch/opsorch-mcp

(+ adapters in the OpsOrch org)

Would love feedback on architecture, adapter model, security concerns, and which integrations you’d want next.