2026-10-08
Day 8: Swift Justice
Lane: Binary reverse engineering. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
binary-reverse-engineering-day-08-swift-network-configuration - Prompt: "Swift binaries speak in protocols and hide their networking in plain sight. Follow the symbols to the URL configuration and see where this one phones home."
What we were given
- One
evidence.zip(31,454,414 bytes, SHA-2562f6bce1ccadc28c17e60783ad22c4e5c3029aef5a2c462803358e8ef5e94f31b). AES encrypted, passwordinfected. - The earlier files again:
d01tod07(C, C++, .NET, Go, Rust, Java, Kotlin). - New today:
d08--variant-000.elf(67,264,576 bytes), a 64-bit Linux program written in Swift.
What was asked
- 80 questions in Single Solve and Time Trial. Mastery had five deliveries, so 400.
- Question types:
- Hashes: SHA-256 of each member.
- Older specimens, new fields today:
- d01: entry point (lowercase hex with
0x) and whether.symtabexists. - d02: whether a symbol matches
/ZTI.*DiscoveryEngine/(typeinfo) or/ZTV.*DiscoveryEngine/(vtable). - d03: config
urland CRC32. - d04 to d07: the XOR key. d04 and d05: the
HCFGversion byte. - d06 and d07: the ZIP compression methods used in the JAR, as a JSON array.
- d08 (new): ELF class as a bit width (the integer 32 or 64), machine, format, architecture, the config fields, and the CRC32.
How it was solved, start to finish
- Ran the driver
run_tm.py: start the attempt, download the zip, runsolve_tm.py, submit. - New fields on old specimens were added to
solve_tm.py, each read straight from the bytes:
- Entry point:
e_entryfrom the ELF header, printed as0x…. .symtab: present in the section name list or not.- Typeinfo / vtable: does the file contain
_ZTI(or__cxxabiv1) /_ZTV. - Compression methods: the set of ZIP method codes over every central-directory entry.
- Example answers:
d01_analysis_entry_point=0x401040,d06_analysis_compression_methods=[0,8],d04_analysis_xor_key_hex=0x51.
- d08 class: the question asked for the bit width, so the solver reads the question text. If it says "bit width", it answers
64instead ofELF64. - d08 config (the new part): the question gave a fixed order:
- First test each literal
HCFGin the file. - Then Base64-decode every run of 100+ Base64 characters.
- Accept the first CRC32-valid
HCFGrecord.
The saved Swift sample has the text HCFG once, but it is not a valid record. The real one is a Base64 string. Re-checked:
[base64 run at file offset 18366768, run length 508]
HCFG version=1 xor_key=0x41 len=368 crc32=0x6a033bfd
section of match: .rodata
JSON: {"campaign":"BR08-000","domain":"node-000.day08.example.invalid","ipv4":"198.51.100.160","url":"https://node-000.day08.example.invalid/api/BR08-000","mutex":"Global\\HuntressBR08_000", ...}
solve_tm.py already does literal first, then Base64 runs (its "step 3" for Swift and Nim), so it found this record.
- Submitted all 80 answers (
answers.json). Result:CORRECT. - Time Trial: a fresh variant, same steps. Result:
CORRECT. - Mastery: five deliveries (400 answers), same steps per delivery. Result:
CORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 2055206eaef175e31045e75a89bd1760 |
| Time Trial | CORRECT | 26ac8723e4fc810269ad4dbe744e2d44 |
| Mastery | CORRECT | d279a05b1cf9ffcf61763b6803a823fa |
Each mode folder has the raw start-response.json, answers.json, submit-response.json and final-status-response.json.
Reproduce
Offline, from this folder (no Game Master needed):
mkdir -p /tmp/d08/deliveries && cp single-solve/evidence.zip /tmp/d08/deliveries/delivery-1.zip
python3 -I ~/.config/huntress/solvers/solve_tm.py single-solve/start-response.json /tmp/d08/deliveries /tmp/d08/answers.json
Checked on 2026-11-01: this gives the same 80 answers as answers.json. It only parses the samples and never runs them.
Live driver:
python3 -I ~/.config/huntress/solvers/run_tm.py binary-reverse-engineering-day-08-swift-network-configuration single_solve /tmp/d08
This needs a Game Master token in ~/.config/huntress/gm-token.
Day 8: What's in the Box?
Lane: Metadata forensics. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
metadata-forensics-day-08-whats-in-the-box - Prompt: "The count was only the beginning. Inventory every stored path, preserve directory context, and identify the archive whose names reveal an impossible provenance claim."
What we were given
- Single Solve:
(evidence.zipmetadata-day-08-evidence.zip, 3,308 bytes, SHA-25612f42a7fcd717ee6105698873b38b6144ac698b7a6a145dbb2c0a5af7110c9c6). - Time Trial:
metadata-day-08-evidence.zip(3,308 bytes, SHA-25643b93b5abe3dd8d91b30ff7525141aa09caae77bd75222ac03eebb4a48d30a98). - Mastery:
metadata-forensics-day-08-whats-in-the-box-mastery.zip(17,252 bytes, SHA-2569e293f86aca6dcc872e214e1a0fb0a10d807ba9d7ced1b0a6bb164cedbf3c529). It holds five inner zips,metadata-day-08-evidence-01.zipto-05.zip, one per set. - Every evidence zip holds the same five file names:
evidence-note.txt,camera-001.jpg,panel-001.png,document-001.pdfandevidence-drop-001.zip.
What was asked
- Single Solve: 12 questions. Time Trial: 12 questions. Mastery: 60 questions (5 sets × 12).
- Question types:
- File identity (evidence note): filename, size, MD5, SHA-1, SHA-256, first 16 bytes as hex, UTF-8 BOM present.
- Headers: JPEG SOF size, PNG IHDR size, PDF header version, ZIP EOCD total entries.
- New today:
zip.central_directory_member_namesforevidence-drop-001.zip. Format: "compact valid JSON array with exact values and order".
How it was solved, start to finish
- Read the new question. It wants every stored path in the inner zip, with folder names kept.
- Added a handler to
run_meta.py. It reads the names from the zip central directory with Python'szipfile. If that fails, it falls back to listing the files that 7z extracts.
if 'central-directory-member-names' in aid:
try: names = zipfile.ZipFile(io.BytesIO(fb)).namelist()
except Exception: names = list_names(fb)
return json.dumps(names, separators=(',', ':'))
On Day 8 the names went out in central-directory order. On Day 23 the handler was changed to sort them (sorted(names)), after the grader showed it wants sorted lists.
- Ran
run_meta.pyfor all three modes. A helper agent ran the driver while other lanes were solved in parallel. Every other field used the handlers from Day 7. - Answers. In all three modes
evidence-drop-001.zipheld the same two paths:
["room-01/item-1.txt","room-02/item-1.txt"]
Other Single Solve answers, from answers.json: evidence-note-size-bytes = 33, d04-jpeg-001-jpeg-sof-dimensions-pixels = 257x146, d07-zip-001-zip-eocd-total-entries = 2.
- Submitted all answers in one JSON object per mode. The Game Master replied
CORRECTand returned the flag.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 6c24273bdc38292421cdc5ca8bf772ee |
| Time Trial | CORRECT | ec09a287c8ed3875180aaf62e41a0bc1 |
| Mastery | CORRECT | e8a42d101c07d1ed15b7f12a0c846828 |
Each mode folder has answers.json, start-response.json, submit-response.json and final-status-response.json. The Single Solve folder also has the zip. No screenshots were captured.
Reproduce
python3 -I ~/.config/huntress/solvers/run_meta.py metadata-forensics-day-08-whats-in-the-box single_solve /tmp/d08
This needs a Game Master token in ~/.config/huntress/gm-token. For time_trial and mastery it calls op=retry first, which spends one of the unique attempts. It only parses the files.
Day 8: Cleartext Session
Lane: Mobile (Android). Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
mobile-android-day-08-cleartext-session - Prompt: "Recover a synthetic session from allowed cleartext traffic and replay it against the backend."
What we were given
- One Android APK per case. Inside:
AndroidManifest.xml,classes.dex,assets/huntress_puzzle.json, andMETA-INF/signing files. - Single Solve:
cleartext_session.apk, caseMOB-08-9DC6C997E6(12,704 bytes, SHA-25660fb542da7a218f5ea4ccd7a9c6015b6349753b46aae9093729ca8be0020bad8). Its puzzle is saved assingle-solve/huntress_puzzle.json. - Time Trial: one APK (case
MOB-08-36AD857EA5). Mastery: five APKs. These APKs were not saved in the repo.
What was asked
- One question per case: return the exact backend answer code for that case.
- Answer format:
ANS-plus exactly 16 uppercase hex characters. - Answer IDs:
answerfor Single Solve and Time Trial.set1.answertoset5.answerfor Mastery. - Mastery grew from three cases to five today.
How it was solved, start to finish
The app was never installed or run.
- Started the Single Solve and downloaded the APK.
- Read the puzzle asset. The codec is
reverse(also used on Day 2).
{"schema":"huntress.mobile.puzzle.v1","day":8,"codec":"reverse","data":"6667D24E8BC6DE39-SNA"}
- Decoded it. Reverse the string:
import json
p = json.load(open('single-solve/huntress_puzzle.json'))
print(p['data'][::-1])
Output: ANS-93ED6CB8E42D7666.
- Ran the lane driver for all three modes. A helper agent ran it once per mode:
python3 -I ~/.config/huntress/solvers/run_mode.py mobile-android-day-08-cleartext-session <mode> mobile_puzzle <workdir>
The driver downloads each APK, reads its puzzle, decodes it by its codec, and submits all answers at once.
- Results: Single Solve (
ANS-93ED6CB8E42D7666), Time Trial (ANS-608057A4A2A9F3AD) and all five Mastery answers came backCORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | d566a211d4967b562dca46cd0cde9a68 |
| Time Trial | CORRECT | 387c15276d932edefa63151cd25ac2a7 |
| Mastery | CORRECT | cd31514b0bb0c09bfa2e3eb763cba6eb |
Each mode folder has answers.json, start-response.json, submit-response.json and final-status-response.json. No screenshots were taken.
Reproduce
Offline: run the Python snippet above from this folder.
A fresh run with the driver:
python3 -I ~/.config/huntress/solvers/run_mode.py mobile-android-day-08-cleartext-session single_solve mobile_puzzle /tmp/d08
This needs a Game Master token in ~/.config/huntress/gm-token. It only unzips the APK and reads the asset. It never installs or runs the app.
Day 8: Lost in Translation
Lane: Script analysis. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
script-analysis-day-08-cross-language-iocs - Prompt: "The same indicators show up in every language the campaign speaks, each with its own quoting, escaping, and encoding accent. Normalize them to their exact form and find the one that doesn't match its siblings."
What we were given
- One
evidence.zip(76,687 bytes, SHA-25682856ed6ca80c3db2e2528c4dbee30874b9643c8d6f5652b7761e7439f8548fe). Password:infected. - The Day 7 set, two more advanced Bash files (
array-slicing.sh,associative-arrays.sh), and two new families: - host (4 files): a script inside a host container:
host-wsf-jscript-jscript.jshost-wsf-mixed-vbscript.vbshost-hta-jscript-javascript.jshost-php-webshell-php.php- polyglot (8 PHP files):
polyglot-php-<role>.php, one per role (benign utility, stager, dropper, downloader, uploader, launcher, persistence descriptor, staged chain). - Every host and polyglot file carries a comment marker,
HUNTRESS_ANALYSIS_V1:<ENC>-<data>. Real starts of markers from this archive:
HUNTRESS_ANALYSIS_V1:B64-dj0zfGxhbmd1YWdlPWpzY3JpcHR8cm9sZT1zdGFn...
HUNTRESS_ANALYSIS_V1:HEX-763d337c6c616e67756167653d7068707c726f6c...
HUNTRESS_ANALYSIS_V1:HB64-646a307a66477868626d64315957646c50585a69...
HUNTRESS_ANALYSIS_V1:FRAG-763d337c6c616e6775616765.3d6a61766173637...
What was asked
- 615 questions: the Day 7 families, 44 host questions and 88 polyglot questions (11 per file).
- Host and polyglot questions per file:
- Role and ATT&CK ID.
- Adapter, campaign ID, and the URL or file path.
- Host profile (
standalone,wsf_jscript,wsf_mixed,hta_jscript,php_webshell). - Obfuscation (
plain_hex,base64,hex_base64,fragmented_hex). - Stage profile (
single_descriptor,two_stage_download_execute,three_stage_decode_write_launch,cross_language_handoff). - Next language, stage count and steps.
How it was solved, start to finish
- Started the untimed Single Solve and downloaded the archive.
- Looked at the new files. The marker comes in four encodings:
B64: Base64.HEX: hex.HB64: hex that decodes to Base64 text, which then decodes to the config.FRAG: hex split by dots.
- Wrote
hostpgdec.py. It finds the marker with a regex, decodes it by its tag, and splits the result on|and=:
m = re.search(r'HUNTRESS_ANALYSIS_V1:([A-Z0-9]+)-([^\s"\'<]+)', text)
enc, data = m.groups()
raw = {'B64': b64, 'HEX': unhex, 'FRAG': unhex, # unhex() drops the dots
'HB64': lambda d: b64(unhex(d).decode())}[enc](data)
cfg = dict(p.split('=', 1) for p in raw.decode().split('|'))
- No guessing is needed for these families. The decoded config already holds the role, ATT&CK ID and every profile. Example,
host-wsf-jscript-jscript.js:
v=3|language=jscript|role=staged_chain|verdict=suspicious|attack=T1059|adapter=operation_graph|host_profile=wsf_jscript|obfuscation_profile=base64|stage_profile=two_stage_download_execute|next_language=jscript|stage_count=2|steps=acquire_remote,dispatch_process|campaign_id=JSCRIPT-164f9679|url=https://jscript-164f967902.example.invalid/stage|file_path=/tmp/huntress-jscript/.164f967902e48f53.stage
The only change is steps: the commas become >.
- Added host and polyglot to
solve_expansion.py. The question'sartifactfield gives the file. The answer suffix (-role,-host-profile,-next-language, ...) gives the config key. - Submitted. All 615 answers were accepted, and the Single Solve was
CORRECT. Real answers for one polyglot file:
{"polyglot-php-downloader-role": "downloader", "polyglot-php-downloader-attack": "T1105",
"polyglot-php-downloader-adapter": "http_client", "polyglot-php-downloader-campaign": "PHP-e379b654",
"polyglot-php-downloader-steps": "profile_host>decode_configuration>dispatch_language",
"polyglot-php-downloader-obfuscation": "fragmented_hex",
"polyglot-php-downloader-stage-profile": "cross_language_handoff",
"polyglot-php-downloader-next-language": "vbscript"}
- Time Trial and Mastery ran with
run_expansion.py. Both had no deadline (run_for_time: false). Mastery had 3,075 questions in five sets, from one bundle with an inner zip per set. Both wereCORRECT. - While the script module was being written, the other four lanes of the day were solved in parallel by sub-agents.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | eadddef10b2f618ecfcb050ced73d3bf |
| Time Trial | CORRECT | d5ae3da071aedc6ccf6439812d284b99 |
| Mastery | CORRECT | e32767c534426c6bd7912b4602b5d7db |
Each mode folder has answers.json and the Game Master JSON. The Single Solve folder also keeps evidence.zip. No screenshots were taken.
Reproduce
Offline, from the repo root:
mkdir -p /tmp/d08/ex
7z x -pinfected -o/tmp/d08/ex repos/day-08-script-analysis-lost-in-translation/single-solve/evidence.zip
python3 -I tools/solve_expansion.py /tmp/d08/ex \
repos/day-08-script-analysis-lost-in-translation/single-solve/start-response.json /tmp/d08/answers.json
This rebuilds all 615 saved answers (re-checked on 2026-11-01). It only parses the files.
On the day, run_expansion.py drove the live modes (it needs a Game Master token). Since Day 23 that driver sends any question set with polyglot- IDs to solve_day23.py, which does not handle this day's host- and resource- IDs. For a fresh attempt of this format, extract its archive and run tools/solve_expansion.py on it as shown above.
Day 8: Entity Relations
Lane: Warmups. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
warmups-day-08 - Prompt: "The web has a vocabulary for unsafe characters. Speak its language."
- New encoding: HTML entities
What we were given
- One blob file per mode (Single Solve and Time Trial), or a zip of multiple blobs (Mastery).
- Each blob has numbered lines (
NN<TAB>fragment). Each line is independently encoded. - The decoded text ends with a slug like
word-word-word-NNto submit.
How it was solved
- Decode the blob. HTML character references (
&#NNN;or&#xHH;). Unescape to get plaintext. - Mastery blobs use multiple encodings from Days 1-8. The solver (
warmup_decode.py) tries every known decoder on each line and picks the one producing printable text. - Concatenate all decoded lines. The text contains a message ending in the answer slug.
- Submit the slug to the Game Master.
Results
- Single Solve:
CORRECT. slug:jovial-rocket-marble-82. Flag:5f0688047819dea6c1b6406c2cab4320. - Time Trial:
CORRECT. slug:bright-blanket-ivy-12. Flag:6327602f7f64e77a5dfe482e7fdf1573. - Mastery:
CORRECT. 38 slugs. Flag:450f6814d81baa297fff23dd71bbd677.