provider paths: an answer code hex-encoded inside the puzzle asset

huntress ctf 2026, day 6 · mobile android · all three modes

tools used: android apk toolkit and timed-attempt pipeline.

as on days 1-5, the apk carried assets/huntress_puzzle.json and the answer was hidden in it. the prompt told a content-provider story (an overbroad provider path, records correlated with earlier intent evidence), but the graded field was again the backend answer code, and day 6's codec was hex.

the key insight: puzzle codec names are aliases onto the shared decoding registry, which already had hex, so no code changed. the story in the manifest is flavor; the puzzle asset is the whole problem, and the expected answer shape is the validator that decides when decoding is done.

challenge shape

one field, answer, per apk: ANS- plus sixteen uppercase hex digits. the single and the time trial carried one apk, mastery three.

artifact anatomy

{"schema": "huntress.mobile.puzzle.v1", "day": 6, "codec": "hex",
 "data": "414e532d34364132344243444145354336453142"}

forty hex digits are twenty bytes, which is exactly the length of ANS- plus sixteen characters. the answer is therefore the ascii text of the decoded bytes, not the hex string itself, even though the answer's own tail is hex too. the stored digits are lowercase and always start 414e532d (ANS-); the answer comes out uppercase because the code's own letters are uppercase ascii.

answer = bytes.fromhex(puzzle["data"]).decode("ascii")

extraction method

the codec field may be a chain of steps (base64+reverse, hex|rot13), applied left to right. each name is an alias onto the decoding registry, and keyed steps (xor, add, rc4, vigenere) take the puzzle's key:

CODEC_ALIASES = {"base64": "b64", "hex": "hex", "reverse": "reverse", ...}

out = puzzle["data"]                      # simplified: apk._chain keeps every reading
for step in _steps(puzzle["codec"]):      # aliases resolved, chain split
    out = decode.TRANSFORMS[step](out)    # keyed and byte steps branch here

validation and answer formatting

the decoded text must match ANS-[0-9A-F]{16} exactly. if the declared codec does not produce that shape, the toolkit does not trust the declared codec: it peels the data through the generic transform search with the answer shape as the goal. here the declared codec was honest and the first decode matched.

failed approaches and pivots

none.

automation across modes

the solver opens every apk in the set, reads the puzzle asset, decodes it and checks the shape, so mastery's three apks were three runs of the same function.

lessons

solve times

public scoreboard · day 6 finish: 1st

  • 0:01.9singlefirst blood, +2.0 s
  • 0:04.0time trialfirst blood, +5.4 s
  • 0:06.6masteryfirst blood, +6.6 s
  • #1team day finishof 182 teams that cleared it, in 1:45.7
  • #2single orderof the day's 5, 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 2s 1 0
time trial 2s 1 0
mastery 2s 1 0

the clock

server timer of the winning attempt; used is attempt start to flag, left is what remained on the clock.

mode time available time used time left
time trial 6.0s 1.0s 5.0s
mastery 6.0s 1.0s 5.0s

where we got stuck

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

{generated by a clanker}