In a nutshell Wireshark captures the actual packets flowing over a network and shows them to you, decoded layer by layer. Where Nmap asks "what's there?", Wireshark answers "what exactly is being said?" — letting you see a login sent in the clear, a protocol misbehaving, or malware phoning home.
What it does — the layers peeled open
Wireshark records every frame the card sees and decodes each through the layers (see OSI & TCP/IP models):
You get the raw truth of what crossed the wire: MAC (L2), IPs (L3), ports & TCP (L4), and the actual content (L7).
Analogy — a transcript of every conversation in a room. Nmap tells you which phones are on. Wireshark hands you the word-for-word transcript of every call — who spoke to whom, and exactly what they said.
Staying afloat — filters
A busy capture is thousands of packets a second. Display filters narrow the flood:
ip.addr == 10.10.10.5 only traffic to/from this host
tcp.port == 80 only web traffic
http only HTTP
dns only DNS queries/answers
http.request.method == "POST" only form submissions
Then right-click a packet → Follow TCP Stream to reassemble a whole conversation into one readable block.
What you actually hunt for
- Clear-text credentials — plain HTTP, FTP, Telnet send usernames and passwords unencrypted; they sit right there in the bytes. The visceral lesson in why HTTPS matters.
- What a host talks to — filter by a suspicious machine; is it beaconing to an unknown server at regular intervals (malware)?
- Protocol behaviour — the TCP handshake, retransmissions, a stalling TLS handshake, DNS failing — gold for attacking and fixing alike.