if a breach falls in a log: an export missing from its audit trail

huntress ctf 2026, day 28 · web exploitation · all three modes

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

the target offered an export operation and a separate status view. the question was whether the export appeared in the audit record. the decisive evidence was the relationship between the current export id, the status report's declared fields, and the events it listed.

challenge shape

single and time trial assigned one temporary target each. mastery assigned five targets that could belong to different web families. each answer was an exact proof recovered from its corresponding target.

artifact anatomy

the first page led to a json clue document. its feature route required POST; a GET returned post_required. an empty json post created an export and returned an export id. a secondary route returned a report describing the request method and fields, together with event records. these are target-local observations: ids and paths can change between attempts.

extraction method

the solver reads the clue, posts to its feature route, and then requests the clue's secondary route. it checks the report's method and field list before constructing the follow-up request. it pairs an export id from the current target with its observed event and the report's missing event. a response must actually contain one proof before the candidate is accepted.

clues = read_clues(target)
export = post_json(clues["feature_route"], {})
status = get(clues["secondary_route"])
if report_supports_audit_replay(status["report"]):
    follow_up = build_from_current_export_and_events(export, status)
    proof = post_json(clues["secondary_route"], follow_up)

validation and answer formatting

every route is resolved against the assigned origin. the follow-up uses the export and events observed on that target; a successful http status alone is insufficient. in mastery, a proof stays paired with the target that produced it. submitted results are [redacted].

failed approaches and pivots

the first request treated the feature as a read route and received the method error. two successive recipe revisions captured useful responses but stopped before the complete export-to-report-to-event sequence, so neither yielded a proof. the final revision carried all three observations into one request flow.

automation across modes

the capture classifies each target independently. one target can use this export flow while another uses an earlier family. the same current-target contract is read again for time trial and every mastery target.

lessons

solve times

public scoreboard · day 28 finish: below top 10

  • 3:24.7single#36 of 147 solvers
  • 3:33.7time trial#28 of 144 solvers
  • 3:45.0mastery#12 of 130 solvers
  • #49team day finishof 122 teams that cleared it, in 31:21.7
  • #4single orderof the day's 7, by our flag time

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

work began when the release assigned the target; the recorded flags were at +205.4s, +214.5s, and +225.9s. the timed rows measure from the preceding mode's flag.

mode work to flag attempts wrong submits
single 3m25.4s 1 0
time trial 9.1s 1 0
mastery 11.4s 1 0

the clock

the winning assignments recorded their start and deadline. time used is assignment start to flag; small rounding differences reflect timestamp precision.

mode time available time used time left
time trial 125.5s 8.2s 117.3s
mastery 123.8s 10.7s 113.1s

where we got stuck

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

{generated by a clanker}