Infrastructure MCP OPS Bundle

Managing complex infrastructure across Linux hosts, Docker, Proxmox, and Kubernetes often leads to "blind" troubleshooting and risky, unverified changes. This skill package integrates directly with an Infrastructure MCP Server to provide read-only fleet monitoring, evidence-based incident triage, and gated remediation workflows.

What it does

• Fleet Health Briefing generates prioritized status reports with explicit coverage accounting for servers and services.
• Evidence-Based Triage diagnoses outages by correlating signals across network, host, container, and platform layers.
• Gated Remediation executes container restarts or VM actions using a strict dry-run, approval, and independent verification cycle.
• Hypothesis Ranking produces ranked diagnostic possibilities with cited evidence, distinguishing between confirmed signals and hypotheses.
• Safety Enforcement prevents unauthorized mutations by requiring explicit user confirmation and multi-tool verification.

How it works

1. Scope and Discover by calling list_servers to map the environment and identify candidate targets by tags or names.
2. Collect Evidence using broad-to-narrow read-only tools to gather logs, metrics, and health states across the infrastructure stack.
3. Plan and Preview by presenting a specific action plan and running dry_run: true to show the exact command impact.
4. Execute and Verify by applying approved changes with confirm: true followed by independent post-action health checks.

Frameworks & tools

Requires a running Infrastructure MCP Server. Supports environments involving Linux hosts, Docker, Kubernetes, Proxmox, Grafana, and Cisco networking devices.

Why this beats prompting it yourself

Generic prompts often ignore the "coverage gap" problem, assuming a lack of error means health. These skills enforce a strict evidence discipline that cites specific tools and timestamps while preventing accidental destructive commands through mandatory dry-runs and independent verification steps.

Use cases

• Generating a shift-handover brief that accounts for all unchecked or failing targets.
• Triaging a "service unreachable" alert by correlating host health with container logs.
• Safely restarting a production container with a pre-defined rollback and verification plan.
• Mapping dependencies before taking a Proxmox VM snapshot to assess the blast radius.

Known limitations

Requires an active Infrastructure MCP Server connection. Does not support shell execution, custom script injection, or mutating actions outside of the specific supported toolset.