network-engineer

Diagnose network connectivity, latency, packet loss, DNS, and AWS VPC issues using decision trees and CLI commands.

3|1|Updated Feb 14, 2026
One-click install
npx skills add https://github.com/LiboMa/agenticops-chat --skill network-engineer-liboma
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: network-engineer
Source: https://github.com/LiboMa/agenticops-chat/tree/main/skills/network-engineer
Command: npx skills add https://github.com/LiboMa/agenticops-chat --skill network-engineer-liboma

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Network outages and performance degradation are hard to isolate because failures can occur at any layer — interface, routing, firewall, DNS, MTU, or cloud security rules. This Skill provides CCIE-level troubleshooting decision trees and command references so an agent can systematically isolate root causes instead of guessing. ## Core Features & Use Cases - Guided Decision Trees: Step-by-step escalation paths for connectivity failures, latency, packet loss, and DNS problems, with concrete commands like mtr, ss -ti, dig, and tcpdump at each branch. - AWS VPC Networking Diagnostics: Systematic 8-step checks covering Security Groups, NACLs, route tables, NAT Gateways, Transit Gateways, VPC peering, and load balancer health checks. - Deep Reference Material: Detailed references on firewall analysis (iptables, conntrack, SG/NACL semantics), routing troubleshooting (BGP, OSPF, TGW, asymmetric routing), and TCP internals (window sizing, retransmissions, keepalive, MSS/MTU). - Use Case: When users report intermittent timeouts to an internal service, walk the connectivity decision tree, verify SG/NACL rules on both subnets, check conntrack table utilization, and pinpoint whether the issue is a stateless NACL blocking ephemeral ports. ## Quick Start Use the network-engineer skill to diagnose why instances in subnet A cannot reach the database on port 5432 in subnet B.

Frequently Asked Questions about network-engineer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I troubleshoot network connectivity issues on Linux?▼

Follow a layered approach: verify the interface is up with ip addr show, check the default gateway with ip route show, ping the gateway, then run mtr to the target to find where packets drop. Finish by testing DNS with dig and the specific port with nc -zv.

How do I debug AWS Security Group and NACL connectivity problems?▼

Check the source SG outbound rules, source subnet NACL outbound, route tables, destination NACL inbound, and destination SG inbound in order. Remember NACLs are stateless, so ephemeral ports 1024-65535 must be allowed for return traffic, while Security Groups are stateful.

Why does my connection work to the IP but not the port?▼

This indicates a transport-layer block rather than a routing issue. Test with nc -zv target_ip port, then check Security Groups, NACLs, and local iptables rules, and confirm the service is actually listening with ss -tuln on the target.

How do I find the path MTU between two hosts?▼

Use ping with the don't-fragment flag: ping -M do -s 1472 -c 1 target tests a 1500-byte path. If it fails, binary search with smaller sizes like 1400 or 1300 until it succeeds, then add 28 bytes for IP and ICMP headers.

What causes high TCP retransmission rates?▼

High retransmissions indicate real packet loss in the path, congestion, or interface-level drops. Check ss -ti for per-connection retransmits, ip -s link for RX/TX errors, and mtr for loss at intermediate hops; a ratio above 1% of segments is concerning.

Why is my BGP session stuck in Active or OpenSent state?▼

Stuck in Active means the TCP connection on port 179 is failing, usually due to firewall rules or a wrong peer IP. Stuck in OpenSent indicates the OPEN message was rejected, typically from an AS number mismatch or router-id conflict.