MCP Incident Intel
brianpelow/mcp-incident-intel
Integracao PagerDuty + Dynatrace + runbooks para resposta automatizada de incidentes
Ver no GitHubServidor MCP que integra PagerDuty, Dynatrace e runbooks para resposta a incidentes orientada por IA. Fornece contexto unificado de incidentes, permitindo que agentes LLM façam triagem, correlação e início de remediação sem trocar entre ferramentas.
Instalação
npx convez-ai brianpelow/mcp-incident-intel1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Triagem e correlação automatizada de incidentes em tempo real
- Aceleração de MTTR (Mean Time To Resolution) em ambientes críticos
- Contexto unificado para agentes IA em operações de SRE
Veredito

Solução madura e bem-estruturada para times de SRE/Platform Engineering que usam PagerDuty e Dynatrace. Reduz significativamente tempo de resposta a incidentes ao centralizar observabilidade e runbooks num único ponto de acesso para Claude.
README
A tradução para o português deste README ainda está na fila.
mcp-incident-intel
MCP server wiring PagerDuty, Dynatrace, and runbook context for AI-driven incident response.
Overview
mcp-incident-intel is a Model Context Protocol server that gives AI agents
full incident context in a single, standardized interface. It wires together
PagerDuty incident data, Dynatrace observability signals, and operational
runbooks so that LLM-powered agents can triage, correlate, and begin
remediation without switching between tools.
Built for SRE and platform engineering teams in regulated financial services and manufacturing where mean time to resolution is a compliance metric.
Tools exposed
| Tool | Description |
|---|---|
get_active_incidents | Fetch open PagerDuty incidents with severity and assignee |
get_dynatrace_problems | Fetch open Dynatrace problem cards with root cause |
get_service_topology | Get upstream/downstream dependencies for a service |
get_slo_status | Fetch SLO burn rate and remaining error budget |
get_runbook | Retrieve runbook content for a given alert name |
correlate_incident | Cross-reference a PagerDuty incident with Dynatrace problems |
Quick start
pip install mcp-incident-intel
export PAGERDUTY_TOKEN=your_pagerduty_token
export DYNATRACE_URL=https://your-env.live.dynatrace.com
export DYNATRACE_TOKEN=your_dynatrace_api_token
mcp-incident-intel
Configuration
| Variable | Description | Required |
|---|---|---|
PAGERDUTY_TOKEN | PagerDuty API token | Yes |
DYNATRACE_URL | Dynatrace environment URL | Yes |
DYNATRACE_TOKEN | Dynatrace API token | Yes |
RUNBOOK_DIR | Path to runbook markdown files | No |
Contributing
See CONTRIBUTING.md.
License
Apache 2.0 — see LICENSE.