Skip to main content
Glama
Infrastructure MCP Server

Infrastructure MCP Server

by berkeleyo

Infrastructure MCP Server

An independently authored, read-only Model Context Protocol server by berkeleyo. It exposes allowlisted multi-vendor infrastructure discovery, desired-state drift analysis, validation, and bounded network diagnostics to compatible AI clients. Provider discovery is delegated to the commit-pinned multi-vendor-infrastructure-automation package rather than duplicated here.

Offline inventory and mocked network boundaries are tested. No live cloud or firewall connectivity is claimed or required.

Architecture

flowchart LR
  C[MCP client] -->|stdio JSON-RPC| S[MCP SDK server]
  S --> A[Authorisation and typed validation]
  A --> I[Inventory service]
  A --> D[Bounded diagnostics]
  I --> P1[Project 1 package]
  P1 --> F[Offline fixtures]
  P1 --> L[Authorised read-only providers]
  D --> PUB[Allowlisted public destinations]
  S --> AUD[Redacted audit events on stderr]

Related MCP server: emfirge

Tools

Tool

Purpose

infrastructure_health

Configuration readiness, explicitly not live connectivity

discover_infrastructure

Normalized inventory for an allowlisted profile

detect_infrastructure_drift

Deterministic desired/actual comparison

validate_infrastructure_config

Project 1 schema validation without live calls

diagnose_dns

Bounded A, AAAA, CNAME, MX, and TXT lookups

diagnose_tls

Certificate, SAN, validity, cipher, and protocol inspection

diagnose_http

Body-free HEAD/GET status, timing, and safe headers

list_available_capabilities

Configured capabilities without connection claims

Install and run

python3 -m venv .venv
.venv/bin/pip install -e '.[dev]'
INFRA_MCP_CONFIG=examples/server.yaml .venv/bin/infrastructure-mcp-server

Example client configuration:

{
  "mcpServers": {
    "infrastructure": {
      "command": "/absolute/path/infrastructure-mcp-server/.venv/bin/infrastructure-mcp-server",
      "env": {"INFRA_MCP_CONFIG": "/absolute/path/infrastructure-mcp-server/examples/server.yaml"}
    }
  }
}

The host chooses the executable, configuration, environment, credentials, and OS permissions. Tool approval prompts are useful UX but are not the security boundary. AI clients can select only preconfigured aliases—not arbitrary paths, hosts, URLs, or commands.

Run the deterministic offline suite with scripts/verify.sh. See RUNBOOK.md, security, and integration.

Lifecycle

Plan → Build → Test → Deploy → Operate → Improve: define trust boundaries and least privilege; implement typed read-only tools; test protocol and adversarial paths; package without cloud changes; operate with host-controlled profiles and redacted audit events; improve from reviewed evidence.

Known limitations

  • Live providers have not been smoke-tested by this repository.

  • Diagnostics deliberately reject redirects, private/reserved addresses, and arbitrary paths.

  • HTTP and TLS connect to the validated IP while preserving Host/SNI verification, preventing a second hostname resolution. Run the stdio process with network egress controls for defence in depth.

  • Stdio is local only; no remote HTTP transport or multi-user authentication is supplied.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      D
      maintenance
      Enables AI assistants to inspect and audit Azure Landing Zones by inventorying resources, auditing tagging, evaluating policy compliance, and detecting infrastructure drift, all in read-only mode.
      -
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI agents to run read-only AWS security audits and receive normalized, severity-ranked findings about account posture, such as stale access keys, missing MFA, public buckets, open security groups, and CloudTrail coverage.
      11
      26 PyPI
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI clients to perform safe, read-only IT diagnostics and retrieve local runbooks, asset records, and knowledge articles through MCP, with allowlisted network checks and audit logging.
      MIT