query me this: closing the parenthesized search

huntress ctf 2026, day 19 web track · web exploitation · all three modes

tools used: live web target capture and timed-attempt pipeline.

the target exposed a small record-search form. normal queries returned public rows, while a boolean sql injection exposed a restricted row containing the requested value.

the key insight was syntactic: a quote and boolean expression were not enough. the input also had to close a surrounding parenthesis before adding the always-true condition and commenting out the remainder.

challenge shape

the single and time trial each assigned one search target. mastery combined an earlier backup-exposure target with a search target. variants changed the search route, so the solver parsed the actual form rather than assuming a fixed path.

artifact anatomy

the home page contained a get search form whose action was a lookup- or search-like route. the text field also used a search-like name. ordinary input returned matching records. several malformed or incompletely balanced inputs returned no restricted row rather than a useful database error.

the successful url-decoded input had this structure:

%') OR 1=1--

the wildcard extends the original match, the quote closes its string, the right parenthesis closes the grouped expression, the boolean condition widens the result set, and the comment marker discards the remaining query text. the server source was unavailable, so that query shape is an inference from the observed responses rather than a source-level claim.

extraction method

forms were ranked by their action and input names. hidden inputs were retained, while the best non-hidden field received the payload through the form's own method:

fields = [i for i in form.inputs if i.type not in {"hidden", "submit", "file"}]
parameter = rank_search_names(fields)[0]
values = {i.name: i.value for i in form.inputs if i.type == "hidden"}
values[parameter] = "%') OR 1=1--"
request(form.method, form.action, values)

if the page linked directly to a lookup or search route, that same target-local route was a fallback. no unrelated path probing was needed.

validation and answer formatting

the solver treated the appearance of exactly one requested value as success. returning ordinary records alone did not count. as with the other web recipes, the request could not leave the assigned origin and the response was saved before answer extraction.

the submitted result in every mode was [redacted].

failed approaches and pivots

automation across modes

the page itself selected the sql-injection recipe through its form and search vocabulary. heterogeneous mastery targets were captured concurrently, and the offline answer step mapped each capture back to its own question.

lessons

solve times

public scoreboard · day 19 finish: top 3

  • 2:31.9singlefirst blood, +57.4 s
  • 3:32.2time trialfirst blood, +1:10
  • 17:56.2mastery#4 of 167 solvers
  • #3team day finishof 153 teams that cleared it, in 17:56.2
  • #8single orderof the day's 8, by our flag time

from the public scoreboard: 09:00 et release to our flag; the table below counts from our attempt's start.

from when we started working on the mode to its flag.

mode work to flag attempts wrong submits
single 2m31s 1 0
time trial 9s 1 0
mastery 18s 1 0

the clock

server timer of the winning attempt; used is attempt start to flag, left is what remained on the clock. the accepted mastery run was untimed.

mode time available time used time left
time trial 120s 9s 111s

where we got stuck

flags: single, time trial, and mastery [redacted].

{generated by a clanker}