infrastructure safe remediation

Manual infrastructure changes are prone to human error, forgotten dependencies, and verification gaps. This skill enforces a strict safety protocol for container and VM operations, ensuring every action is planned, previewed, and verified independently. It bridges the gap between raw infrastructure tools and safe operational procedures by requiring explicit approval gates and mandatory dry runs.

What it does

• Enforces safety protocols by requiring dry runs and explicit user approval for every mutation.
• Manages container lifecycle specifically for restarting Docker containers through supported MCP tools.
• Controls Proxmox VMs including start, stop, and snapshot operations with dependency awareness.
• Performs independent verification by checking system state after changes using secondary read-only tools.
• Mitigates blast radius by assessing dependencies and risks before executing any destructive action.

How it works

1. Target Discovery: Use read tools to identify the exact container or VM ID and assess its current health.
2. Blast Radius Analysis: Identify what depends on the target and propose a rollback plan.
3. Dry Run Preview: Execute the tool in dry-run mode to show the user exactly what will happen.
4. Human-in-the-loop Approval: Wait for the user to explicitly confirm the target and action details.
5. Execution and Audit: Run the confirmed mutation and verify the outcome against the original plan.

Frameworks & tools

This skill requires a connected Infrastructure MCP Server. It works specifically with Docker environments and Proxmox virtualization nodes, leveraging the infrastructure-mcp specification for command execution.

Why this beats prompting it yourself

Standard prompts often execute commands immediately or ignore the verification phase. This skill embeds a rigid state-machine logic that prevents accidental batching, forces pre-execution checks, and ensures you never lose track of system health after a restart.

Use cases

• Restarting a hanging Docker container without impacting linked microservices.
• Safely stopping a Proxmox VM after verifying no critical workloads are active.
• Creating a pre-deployment snapshot on a specific Proxmox node.
• Validating that a VM successfully reached a running state after a cold boot.

Known limitations

Supports only Docker restarts and Proxmox start, stop, and snapshot actions. Requires an active infrastructure-mcp server; it cannot execute raw shell commands or scripts.