2026-10-31

Day 31: The Reversing Gauntlet (Final Boss)

Lane: Binary reverse engineering. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.

The challenge

What we were given

What was asked

How it was solved, start to finish

  1. The watcher ran the driver first. It did not finish the attempt (single=None), so the lane was flagged for manual work.
  2. Standard fields: solve_tm.py handled about 379 of the 434 fields per variant (grouped by dNN).
  3. Gauntlet fields: a new helper, tools/gauntlet_bin.py, computes the other ~55 fields straight from the extracted members:
  1. Found the last wrong answers with the partial-submit oracle. A first partial submit came back incorrect (saved as partial369.json). The Game Master keeps accepted answers across partial submits, so per-day and per-field submits showed which fields were wrong. Two variant-specific fixes came out of this:
  1. d31 HGAU record: scan back from the end of the d31 ELF for HGAU, check bounds and CRC32, XOR-decode, parse the JSON. The JSON holds config and trace. Single Solve answers (answers.json):
  1. Single Solve and Time Trial: tools/run_gauntlet_bin.py downloads the zip, runs solve_tm.py and gauntlet_bin.py, merges the two answer sets and submits once. Result: CORRECT for both.
  2. Mastery: tools/run_mastery_bin.py downloads all 15 sets in parallel, solves each set with solve_tm.py + gauntlet_bin.py, and submits all 6,510 answers. Result: CORRECT.

Results

ModeResultFlag
Single SolveCORRECT (434/434)10c35879156f4b1b8f1d5e46a3336eb5
Time TrialCORRECT (434/434)15d0a26fad1a5fb177888339f074cb8e
MasteryCORRECT (6,510/6,510)41d2118cc3e03a14f78196e4fe4d55cb

Saved files:

Reproduce

The Day 31 zips were not saved, so an offline re-run is not possible. Live driver for Single Solve:

python3 -I ~/.config/huntress/solvers/run_gauntlet_bin.py binary-reverse-engineering-day-31-reversing-gauntlet single_solve /tmp/d31

Mastery (start-meta.json holds the download capabilities and questions):

python3 -I ~/.config/huntress/solvers/run_mastery_bin.py /tmp/d31m mastery/start-meta.json

Both need a Game Master token in ~/.config/huntress/gm-token. Note: for Time Trial and Mastery, run_gauntlet_bin.py sends op=retry, which starts a fresh attempt. Neither tool runs the samples.

Day 31: Mobile Arena Gauntlet

Lane: Mobile (Android). Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.

The challenge

What we were given

SetArtifactSHA-256
1set-01.zipcf882f950f8204f0522e75dc5d412513c896bb2e82d1165784e263e1034e341b
2set-02.zipc56ae5588ff146f461db7c7ef875df40c325325fd646db8c7fafb1641a221da6
3set-03.zip19bb596490e3759cc995a33d7ba5f630fbed3f6b9175bd06902698f9e74ec032
4set-04.zip7a21f6287cb40aa48ab9fd8c7c5c8b7525a733b8ecf8b4f3ca9392e97ee39823
5set-05.zipb1124b96fcb35c39790c4c7f3cd4eaf42b77d620a019f4e1fe7ac51b0531cd2e
6set-06.zipcce83c58b2f808d1aa2350640dcc9d6476e6de93814e33ba05c9b1ae8b89c43d
7set-07.zip4423ca10f450ef5f57af8a2726f640a7e43f4c256b4001ef8ff010cf90322217
8set-08.zip883bda0ce5fdda15732e62c7e72c148d9615c0c5bcce7e863577740351366cda
9set-09.zipa4b083782eba08407320fb3ccb04c55acd40ebd3b8da7af15c3657320e3d47fc
10set-10.zip3033faa92ffde2b78780ab80baedbb9a68b17b9ecea28fadbce53c865b0f6f1c
11set-11.zip7336bbb1ef03d937e6d85f8f02abf423cce85969f4e9b708a18fe2c9739243fc
12set-12.zip496e59f8d6b84df89e6f6e8254a9a3d8a40c605dacb108fdb774ea3ba763a15b
13set-13.zip377444eac526fba4764eecc68437ca0ddbb175d721ef6d89f3205d1bc13de720
14set-14.zipb95e4ed7863b9030f697bf55bb7e5530274202bbccc79e56486654c994fc526d
15set-15.zipecd700f8b497dc266ccfb71d30453e34d90d3e2e32ba367bc40a60febb82b90a

What was asked

Answer IDCasePackage
set1.answerMOB-02-38E9261315com.huntress.mobile.d02.v38e9261315e4
set2.answerMOB-03-236070EE4Dcom.huntress.mobile.d03.v236070ee4d6c
set3.answerMOB-05-65A0CCE721com.huntress.mobile.d05.v65a0cce721bd
set4.answerMOB-07-DB46927816com.huntress.mobile.d07.vdb46927816e1
set5.answerMOB-10-C91D772815com.huntress.mobile.d10.vc91d772815e1
set6.answerMOB-11-99C82461C3com.huntress.mobile.d11.v99c82461c3ec
set7.answerMOB-14-1B42E6B9ACcom.huntress.mobile.d14.v1b42e6b9ac88
set8.answerMOB-16-214E11B13Fcom.huntress.mobile.d16.v214e11b13ffd
set9.answerMOB-19-2EBDC84E87com.huntress.mobile.d19.v2ebdc84e8719
set10.answerMOB-21-97EDD052B8com.huntress.mobile.d21.v97edd052b8bf
set11.answerMOB-22-EEF290EF65com.huntress.mobile.d22.veef290ef6502
set12.answerMOB-25-932DFBF5E9com.huntress.mobile.d25.v932dfbf5e9c2
set13.answerMOB-26-CBB32BD5F9com.huntress.mobile.d26.vcbb32bd5f985
set14.answerMOB-28-659AFB110Bcom.huntress.mobile.d28.v659afb110b22
set15.answerMOB-31-27E1C3A293com.huntress.mobile.d31.v27e1c3a293c4

How it was solved, start to finish

No app was installed or run.

  1. Single Solve and Time Trial at the drop. The watcher ran run_mode.py at 07:00 local time. Its log line, 14 seconds later:
[OK] mobile-android single=CORRECT(726ef0597767e2149c3fcc6c483ff96e) TT=CORRECT mastery=no-output

The APKs and answers for these two modes were not saved.

  1. Mastery: the teammate session. It resumed the assigned Mastery attempt and downloaded the 15 bundles through the Game Master proxy. The hashes above confirm the saved files.
  2. Evaluated each APK's own code. tools/dex_answer_vm.py converts the DEX instructions for the existing interpreter, including methods in assets/*.dex. tools/smali_analysis.py models the asset reads, reflection, string operations and native methods this bundle uses.
  3. First submit: 14 of 15 accepted. The missing piece was an auxiliary DEX and the reflective loader that pulls it in. (Set 10 is the only bundle with a second DEX, assets/payload.dex.) In the same commit the tools gained:
  1. Second submit: all 15, CORRECT. The teammate session committed this folder at 13:29:59 UTC.

Same day, second session

A session on this PC also ran the Mastery with ~/.config/huntress/solvers/run_mode.py (from 13:27 UTC). The first runs crashed inside smali_analysis.py. The fixes were:

At 13:37 UTC the driver printed:

{"mode": "mastery", "n": 15, "result": "CORRECT", "flag": "75bd285cf827f63931da0a38e62eae6e", "remaining": null, "missing": []}

Results

ModeResultFlag
Single SolveCORRECT726ef0597767e2149c3fcc6c483ff96e
Time TrialCORRECT6af707e71b27050e4c52139ec2deb057
MasteryCORRECT75bd285cf827f63931da0a38e62eae6e

Reproduce

Mastery, with the tools/ copy of the driver:

cd ~/projects/huntress-ctf-writeups/tools
python3 -I run_mode.py mobile-android-day-31-mobile-arena-gauntlet mastery mobile_puzzle /tmp/d31

This copy reaches the Game Master through the CourseStack proxy (http://10.0.0.219/huntress-ctf-proxy). The direct DEX path needs the Python packages androguard and capstone. The older smali path needs Java and baksmali.jar. Both read and evaluate the DEX in Python. Nothing is installed or run on a device.

Day 31: One More Thing

Lane: Warmups. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.

The challenge

What we were given

How it was solved

  1. Decode the blob. Peel layers: base64-decode, gzip-decompress, base32-decode. The innermost result is plaintext.
  2. Mastery blobs use multiple encodings from Days 1-31. The solver (warmup_decode.py) tries every known decoder on each line and picks the one producing printable text.
  3. Concatenate all decoded lines. The text contains a message ending in the answer slug.
  4. Submit the slug to the Game Master.

Results

Day 31 — Full Stack Incident (Fetch-to-Worker Chain + Detection/Repair)

Lane: Web Exploitation · Challenge ID: web-exploitation-day-31-full-stack-incident

> Reconstruct the unsigned fetch-to-worker incident from the request log. Follow the exposed service routes to reproduce it, then submit a detector and repair that block the attack while preserving signed status checks.

What we were given

A live web target with exposed service routes, a request log, and an incident evidence API. This was the Day 31 capstone — it required both reproducing an attack chain and writing a detection/repair rule.

How it was solved

Step 1 — Recon

  1. Opened the target and checked /api/clues for the route map: dispatch, audit, and worker endpoints.
  2. Read /api/requests — the request log. It showed a history of requests including req-102: an unsigned queued fetch targeting a specific token (e.g. 38fa9180e8d4). This was the incident to reproduce.

Step 2 — Reproduce the attack chain

Replayed the unsigned fetch-to-worker chain using the exposed routes:

  1. GET /dispatch-*?target=38fa9180e8d4 — create the fetch event
  2. POST /audit-* with {"token": "38fa9180e8d4"} — dispatch the worker
  3. GET /worker-*?command=cat%20objective — execute the unsigned worker action
  4. GET /api/evidence — confirm the incident was reproduced and read the detector/repair schema

Step 3 — Write the detection and repair rule

The /api/evidence endpoint showed the event history and described the required response format. The detector and repair needed to:

Submitted the detector/repair JSON to /api/response. The target validated:

All checks passed — the proof was returned.

Mastery (8 targets)

Targets 1–7 were standard web exploitation types (SSTI, mass assignment, reflected XSS, session fixation, auth bypass, backup exposure, SQLi, IDOR, LFI, stored XSS) solved by web_solve.py. Target 8 was another instance of the fetch-to-worker incident.

Results

ModeTargetsResult
Single Solve1CORRECT
Time Trial1 (timed, 16-player pool)CORRECT
Mastery8 (timed, 8-player pool)CORRECT