Infrastructure Operations Skill

Comprehensive infrastructure health monitoring and container operations management.

Overview

This skill consolidates infrastructure operations including:

• •Service Health Checks: Validate running services and containers
• •Container Discovery: Document and track Docker containers
• •System Diagnostics: Health validation

Value: Unified approach to infrastructure monitoring with consistent status reporting and issue tracking.

Quick Actions

Infrastructure Monitoring Workflow

INFRASTRUCTURE MONITORING
=========================

QUICK CHECK (single service)
  docker inspect <container>
    - Container running status
    - Recent logs (docker logs)
    - Configuration verification
    - Issue storage (Memory MCP if problems found)

FULL CHECK (regular)
  /health-report
    - Docker container status
    - System resource usage
    - Service connectivity
    - Generate report with pass/warn/fail counts

DISCOVERY (new containers)
  docker inspect <name>
    - Container inspection
    - Configuration discovery
    - Documentation creation
    - Registry update (paths-registry.yaml)

Tool Priority

Always use MCP tools first, fallback to bash if MCP fails.

Docker Operations

1. •MCP tools if Docker MCP is configured
2. •Fallback: docker ps, docker logs, docker inspect

SSH Operations (for remote checks)

1. •mcp__ssh__runRemoteCommand if SSH MCP is configured
2. •Fallback: ssh <host> "<command>"

Health Status Reporting

Use consistent severity indicators across all checks:

Thresholds

System Load: Normal < 2.0 | Warning 2.0-3.0 | Critical > 3.0 Memory/Disk: Normal < 80% | Warning 80-90% | Critical > 90%

Memory MCP Storage Pattern

Only store when issues are found.

Storage Pattern for Issues

Entity: "Issue: [Service] [Issue Type]"
EntityType: "Infrastructure Issue"
Observations:
  - Date: [date]
  - Status: [degraded/down]
  - Symptoms: [list]
  - Log errors: [key errors]
  - Severity: [blocker/high/medium/low]
Relations:
  - affects -> [service/container]
  - caused_by -> [root cause if known]

Example - Container Restart Loop

mcp__mcp-gateway__create_entities([{
  name: "Issue: my-service Restart Loop",
  entityType: "Infrastructure Issue",
  observations: [
    "Date: 2026-01-16",
    "Status: Container restarting every 5 seconds",
    "Symptom: Permission denied error",
    "Severity: High"
  ]
}])

mcp__mcp-gateway__create_relations([{
  from: "Issue: my-service Restart Loop",
  to: "my-service",
  relationType: "affects"
}])

Integration Points

With Session Management

• •Health check results can be noted in session-state.md
• •Critical issues should be added to current-priorities.md

With Orchestration

• •Large infrastructure fixes may trigger orchestration
• •Use /orchestration:plan "fix [issue]" for complex repairs

With Memory MCP

• •Issues stored for tracking across sessions
• •Patterns and lessons captured for future reference
• •Use search_nodes to find related past issues

Common Workflows

Quick Docker Check

1. docker ps                    # Check container status
2. docker logs <name> --tail 50 # Check recent logs
3. Review any warnings/errors
4. Add critical issues to priorities if found

Full Health Check

1. /health-report               # Run comprehensive check
2. Review pass/warn/fail counts
3. Address any HIGH or CRITICAL items immediately
4. Create orchestration for complex fixes if needed
5. Update session-state.md with health summary

New Container Discovery

1. docker inspect <name>        # Inspect container
2. Document in paths-registry.yaml
3. Create context file if complex service
4. Add to monitoring rotation

Task Metrics

Track token usage, tool counts, and performance for all Task tool (agent/subagent) executions.

Data source: .claude/logs/task-metrics.jsonl (populated by metrics-collector.js SubagentStop hook)

Commands:

• •/metrics summary - Overview: total runs, tokens, success rate
• •/metrics by-agent [name] - Per-agent stats or single agent detail
• •/metrics by-session [name] - Current or named session breakdown
• •/metrics recent [count] - Last N executions table
• •/metrics top-tokens [limit] - Agents ranked by token consumption
• •/metrics cost - Estimated API cost based on token usage

Troubleshooting

Docker not responding?

• •Check Docker daemon: systemctl status docker
• •Check socket permissions
• •Try: docker info

SSH connection failures?

• •Verify SSH keys: ssh-add -l
• •Check host in known_hosts
• •Test manually: ssh <host> "hostname"

Health check timeouts?

• •Increase timeout in command
• •Run individual checks separately
• •Check network connectivity

Related Documentation

Commands

• •@.claude/commands/health-report.md - Health check

Agents

• •@.claude/agents/service-troubleshooter.md - Systematic service diagnosis
• •@.claude/agents/docker-deployer.md - Guided Docker deployment

Patterns

• •@.claude/context/patterns/memory-storage-pattern.md - Memory MCP patterns