Settings

The capture

Drop a .pcap or .pcapng here or choose one

Read inside this tab by your own browser. It is not uploaded, and once this page has loaded it works with the network off.

How it works

How it reads a capture

  • The file. pcap in all four of its magic numbers, big-endian and little, microseconds and nanoseconds; and pcapng, with the per-section byte order and the per-interface timestamp resolution both honoured. A file cut short mid-packet is read up to the cut and says so.
  • The packet. Ethernet with any depth of VLAN tag, Linux cooked capture v1 and v2, raw IP and BSD loopback; then IPv4 or IPv6 with its extension header chain walked; then TCP, UDP or ICMP. A header that runs off the end of the captured bytes is the normal case when a snap length was set, and is handled rather than thrown on.
  • What it says it is. DNS questions and answers, including the name compression pointers, with a budget so a message that points at itself cannot loop. HTTP request lines and headers, only where the version marker is really on the first line. The server name from a TLS client hello, walked field by field rather than searched for.
  • What stood out. Credentials that crossed in the clear. Flows whose connection attempts are too regular to be a person. Protocols that carry their traffic readably. Names that were looked up and do not exist.
  • What it will not do. It does not reassemble TCP streams, so a request split across segments is not read. It does not decrypt anything. The timer check counts TCP connection starts, so a beacon carried over UDP or QUIC does not show up in it. It is not Wireshark and will not replace it - it answers the first ten minutes of questions, on a machine where you cannot install one and must not upload the file.

The guide: How to read a pcap file without Wireshark

Questions people ask about WIRE

Is the capture uploaded anywhere?

No. The file is opened by your own browser with FileReader and parsed in this tab. Nothing is sent to us or to anybody else, and you do not have to take that on trust: click the Private · verify badge in the corner to see a live count of how much of your content has left the page, or turn off your network and load a capture anyway; it still works.

What files does it read?

libpcap (.pcap, .cap) and pcapng (.pcapng), which is what Wireshark and tcpdump write by default. All four pcap magic numbers are handled (big-endian and little-endian, microsecond and nanosecond timestamps) and pcapng sections that change byte order mid-file, and interfaces that use different timestamp resolutions from each other.

Why not just use Wireshark?

Use Wireshark when you can. This is for when you cannot: a machine you are not allowed to install software on, a borrowed laptop, a phone, or a capture that has to be looked at in the next two minutes. It answers the questions asked in the first ten minutes (who talked to whom, what did they ask for, what went across in the clear, is anything on a timer) and it does not pretend to go further.

How does it decide something is beaconing?

It counts connection starts (a TCP SYN without an ACK) for each source, destination and port, and looks at the gaps between them. If there are at least six gaps and the median absolute deviation is under a fifth of the median gap, it says so. Six because four points can look regular by accident; a fifth because the jitter most implants add is ten or twenty per cent, and a person refreshing a page is nowhere near that regular. Both numbers are printed beside the answer so you can disagree with them.

What credentials can it find?

HTTP basic authentication, which is a username and password in base64 and was never encryption; cookies sent over plain HTTP; and USER and PASS on FTP, POP3, IMAP and Telnet. It reads the command rather than matching a pattern anywhere in the packet, so it reports a password where one was actually sent rather than wherever the word appears.

Does it reassemble TCP streams?

No, and that is a real limit. A request spread across several segments is not read, and a credential split across a segment boundary is missed. What it reads is what fits in one packet, which in practice is most request lines and nearly all DNS. If you need stream reassembly you need Wireshark.

Can it read an encrypted connection?

No. It reads the server name out of the TLS client hello, which is sent before the connection is encrypted and is usually the only name you get for an encrypted flow. Everything after that is ciphertext and stays that way.

What does it do with a capture that was cut short?

Reads it up to the cut and tells you. A capture stopped with ctrl-C often ends mid-packet, and a parser that treats that as corruption throws away everything before it. Everything that parsed cleanly is reported; the note at the top says the file ends mid-packet.

Why does it say a packet is a link type it cannot read?

Because guessing would be worse. The link layer decides where every address and port is, so a wrong guess produces confident nonsense rather than an error. It reads Ethernet, Linux cooked capture v1 and v2, raw IP and BSD loopback; anything else is counted and left alone.

How big a capture can it handle?

It is limited by your browser’s memory rather than by anything here, because the whole file is read into the tab. A few hundred megabytes is fine on a desktop; on a phone, less. If a capture is too big to open, cut it down first with editcap or tcpdump -r and open the piece you care about.

Doing this often? The Sovereign Console loads captures, logs and files into one case, pulls every indicator and timestamp, and exports a report. On its own, US$2 a month; with everything else, US$5 a month or US$48 a year.