Skip to content

Analyze

Statistics

The Statistics tab (F7) summarises the selected sessions — or all sessions when nothing is selected — and updates while traffic arrives:

  • request count, bytes sent and received (bodies), first request and last response, clock duration of the sequence, aggregate session time, in-flight and aborted sessions;
  • throughput (requests and response bytes per second over that duration) and the bytes of request and response headers;
  • Durations of the finished sessions: median and mean, p90 / p95 / p99, minimum and maximum, standard deviation;
  • Connection phases (sum): DNS lookup, TCP connect, TLS handshake and waiting for the first byte, each summed with the number of sessions that had it and the average (a selection of more than 20,000 sessions reads the first 20,000 for this);
  • Response Codes;
  • Response Bytes (by Content-Type) as bars;
  • Hosts and Processes.

Statistics for 24 selected sessions: timing, response codes, bytes by content type and hosts

Timeline

The Timeline tab shows the selected sessions (up to 500) as a waterfall on a shared time axis. Each session's bar is split into its phases:

Phase Meaning
Request from client Quena receiving the request from the client
DNS resolving the server's name
Connect opening the TCP connection
TLS the TLS handshake with the server
Send sending the request to the server
Wait (time to first byte) waiting for the first byte of the response
Receive receiving the response

Hover a bar to see the time of each phase; click a row to select that session. The legend lists the phases that occur; phases without timing data (for example DNS, connect and TLS on a reused connection) are left out.

Structure

Structure in the navigator shows the sessions the filters let through as a tree: hosts at the top, then path segments. Each node shows how many sessions it contains, how many of them are errors and how many bytes were received.

  • Expand hosts and folders to drill down; levels load on demand.
  • Click a host or folder to show only its sessions in the list; click it again (or ✕ above the list) for all. Right-click → Select These Sessions selects them instead — then use Statistics, Timeline, Copy, Save or Remove on exactly that part of the traffic.
  • A filter box narrows the tree.

Compare

Select exactly two sessions and choose Compare from the context menu. Quena shows both sessions — request line, headers and bodies (up to 256 KB each) — side by side as a diff.

Compare captures

Tools → Compare Captures… puts two captures side by side: what changed since the last release, why staging works and production does not, what a new build requests that the old one did not.

  1. Load the captures into the list — Keep and load when Quena asks — or use Load archive… in the dialog. The live capture (what was recorded, not imported) is a side too.
  2. Choose Before and After, then Compare. For two hosts — staging against production — tick Ignore host: requests are then paired by method and path only.

Requests are paired by method, host and path; numbers and ids in the path (/users/123, UUIDs, long hex ids), content hashes in file names (main.3f9a2c1b.js, index-B2x9kQ1a.js) and the values of query parameters do not count, and the n-th call of an endpoint meets the n-th on the other side. The result lists

Mark Meaning
~ changed the status, content type, a response header (added, removed, or the value of Content-Type, Cache-Control, Location, CORS and security headers) or the body differs, or the time is more than twice as long or short (and at least 200 ms apart)
+ new only in After
− gone only in Before
= same nothing of the above (hidden unless chosen)

Groups of sessions compare too: select some sessions and right-click → Compare Groups → Use as Before, then select others → Compare with Before. Pair requests by method and path (the default, as above), method and exact URL, or order (the n-th request of each side with the n-th — two runs of the same steps). Also ignore headers takes response headers that should not count (X-Request-Id; X-Build).

Headers that change on every response (Date, ETag, request ids, Set-Cookie …) are ignored; JSON bodies are compared by content, so key order and spacing do not count. Requests that answered successfully before and fail now (an error status, or no answer at all) are counted apart. Click a row to select its session; double-click a changed one for the text comparison of both. Copy as Markdown puts the result into a ticket or a pull request.

In CI: quena-cli diff; for AI agents: the MCP tool compare_captures.

Text Tools

Text Tools (Ctrl/⌘ E, Tools → Text Tools…, or the wand in the toolbar) convert text you paste:

Operation
To Base64 / From Base64
URLEncode / URLDecode
HTML Encode / HTML Decode
To Hex / From Hex
JS String Escape / JS String Unescape
Decode JWT header and payload as formatted JSON
Unix time → Date seconds, milliseconds or microseconds, as ISO and local time
UTF-8 bytes the UTF-8 byte values of the text