2026-10-05
Day 5: Rust Belt
Lane: Binary reverse engineering. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
binary-reverse-engineering-day-05-rust-runtime-shapes - Prompt: "Rust leaves its own fingerprints: mangled symbols with a distinct accent, panic strings that confess, trait objects hiding in plain sight. Learn the accent."
What we were given
- One
evidence.zip(1,916,070 bytes, SHA-2566049e2c8cfdf9de77efbb58b283233edd7101bfb683b45a7f5277e5db4d3e117). AES encrypted, passwordinfected. - The earlier files again:
d01(C),d02(C++),d03(.NET),d04(Go). - New today:
d05--variant-000.elf(3,997,704 bytes), a 64-bit Linux program written in Rust.
What was asked
- 40 questions in Single Solve and Time Trial. Mastery had two deliveries, so 80.
- Question types:
- Hashes: SHA-256 of each member.
- d01 to d04: the same header, structure and config questions as Day 4.
- d05 (new): container format, CPU architecture, the config fields
campaign,domain,mutex,url, 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. - Routine fields for d01–d04 were answered exactly as on Day 4.
- d05 (the new part): the question said to scan each literal
HCFGin the Rust file and accept the first record whose bounds, CRC32 and JSON all check out.
- The saved sample contains the text
HCFGthree times. Only one of them is a valid record. - The solver's
find_hcfg()tries every match and skips the ones that fail the CRC32 or JSON check, so it picks the right one.
Re-checked on the saved sample:
HCFG offset=355376 version=1 xor_key=0x61 len=368 crc32=0x8f37c323
section of match: .rodata
JSON: {"campaign":"BR05-000","domain":"node-000.day05.example.invalid","ipv4":"198.51.100.100","url":"https://node-000.day05.example.invalid/api/BR05-000","mutex":"Global\\HuntressBR05_000", ...}
Answers: format ELF64, architecture x86-64, CRC32 0x8f37c323, campaign BR05-000.
- Submitted all 40 answers (
answers.json). Result:CORRECT. - Time Trial: a fresh variant, same steps. Result:
CORRECT. - Mastery: two deliveries (80 answers), same steps. Result:
CORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 12b9e4c1568b8fa4333f754d39ba4961 |
| Time Trial | CORRECT | 26e9642973f65fd86ae75be16cbf70f9 |
| Mastery | CORRECT | b0157461375296677b36d1253ffaae91 |
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/d05/deliveries && cp single-solve/evidence.zip /tmp/d05/deliveries/delivery-1.zip
python3 -I ~/.config/huntress/solvers/solve_tm.py single-solve/start-response.json /tmp/d05/deliveries /tmp/d05/answers.json
Checked on 2026-11-01: this gives the same 40 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-05-rust-runtime-shapes single_solve /tmp/d05
This needs a Game Master token in ~/.config/huntress/gm-token.
Day 5: Chunk Norris
Lane: Metadata forensics. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
metadata-forensics-day-05-chunk-norris - Prompt: "The PNGs render, but rendering is not proof. Walk their chunks, verify their declared geometry, and locate the image with an unexpected structural passenger."
What we were given
- Single Solve:
(evidence.zipmetadata-day-05-evidence.zip, 367 bytes). It holds(414 bytes).panel-001.png - Time Trial:
metadata-day-05-evidence.zip(378 bytes, SHA-2560e2866300ff0061a9ff3f3df6a181b5964b27bbfa41e3c223129f5638adc5b05) withpanel-002.png. - Mastery: two deliveries, one per set, with
panel-006.pngandpanel-007.png. The bundle is nested: outer zip →evidence-0N.zip→ PNG. - New today: PNG files.
What was asked
- One field type:
png.ihdr_dimensions_pixelsfor a named PNG. - Answer format:
WIDTHxHEIGHT. - Single Solve and Time Trial: 1 question. Mastery: 2 questions (
set1.,set2.). - Timers: the Time Trial had 50 seconds (
14:55:36→14:56:26UTC). Mastery had 80 seconds.
How it was solved, start to finish
The method: read the true size from the PNG IHDR chunk, not from how the image looks.
- Single Solve. Downloaded the zip and extracted
panel-001.png. A PNG starts with an 8-byte signature. The first chunk isIHDR. Its data starts at byte 16: width (4 bytes) then height (4 bytes), both big-endian.
def png_ihdr(b):
assert b[:8] == b'\x89PNG\r\n\x1a\n'
w, h = struct.unpack('>II', b[16:24])
return '%dx%d' % (w, h)
Result: 269x153. Submitted as d05-png-001-png-ihdr-dimensions-pixels. CORRECT.
- Added
png_ihdrto the driverrun_mode.py(built on Day 4) and ran the Time Trial. Answer forpanel-002.png:282x162.CORRECT. - Mastery, first try failed. The bundle was nested (outer zip →
evidence-0N.zip→panel-XXX.png) and the driver did not look inside the inner zips. - Fixed the driver. It now extracts with
7z x -y -pinfected(plain or AES zips) and searches the zip tree, up to 4 levels deep, for the file named in the question (find_file). - Mastery, after the fix. Answers:
panel-006.png=334x198,panel-007.png=347x207.CORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | c6b8d0a1587b0a922cad26398bfc87f6 |
| Time Trial | CORRECT | bdb56dcda63cbfa01f89ceaafed3d074 |
| Mastery | CORRECT | d97d8a80e66afc8ecd63f5539c7c1bea |
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 and the PNG; its saved submit says already_solved because it is a re-run, and the final status shows CORRECT. No screenshots were captured.
Reproduce
python3 -I ~/.config/huntress/solvers/run_mode.py metadata-forensics-day-05-chunk-norris single_solve png_ihdr /tmp/d05
This needs a Game Master token in ~/.config/huntress/gm-token. For time_trial and mastery the driver calls op=retry first, which spends one of the unique attempts. It only parses the image bytes.
Day 5: Backup Spill
Lane: Mobile (Android). Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
mobile-android-day-05-backup-spill - Prompt: "Extract synthetic application state admitted by a flawed backup boundary."
What we were given
- One Android APK per case. Inside:
AndroidManifest.xml,classes.dex,assets/huntress_puzzle.json, andMETA-INF/signing files. - Single Solve:
backup_spill.apk, caseMOB-05-9E90479049(12,704 bytes, SHA-25650345aca4361eb555258e7bbf3afd862d01c64945c52b8e5af8d1498f7f99719). Its puzzle is saved assingle-solve/huntress_puzzle.json. - Time Trial: one APK. Mastery: three 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.answertoset3.answerfor Mastery. - Time limits: Time Trial 50 seconds, Mastery 110 seconds.
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. New codec today:
split. The code is cut into pieces and shuffled.
{"schema":"huntress.mobile.puzzle.v1","day":5,"codec":"split","data":"","parts":["0A353BE7B","ANS-5","709E18"],"order":[2,0,1]}
- Put the pieces back.
order[i]is the output position ofparts[i]:
import json
p = json.load(open('single-solve/huntress_puzzle.json'))
out = [None] * len(p['parts'])
for i, pos in enumerate(p['order']):
out[pos] = p['parts'][i]
print(''.join(out))
Output: ANS-5709E180A353BE7B.
- Submitted it under
answer. The Game Master repliedCORRECT. - Taught the driver. Added the same
splitlogic tomobile_puzzle()inrun_mode.py. - Time Trial and Mastery: ran
run_mode.pyon both. Both came backCORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 33a15ffa668ec78e55c99f54d9d94e20 |
| Time Trial | CORRECT | 04d7174f5a3dd173a29381dabd1b116f |
| Mastery | CORRECT | 8bbc56e16286fca39182c6887d299501 |
Each mode folder has answers.json, start-response.json, submit-response.json and final-status-response.json. The Single Solve submit-response.json says already_solved because it is a resubmit. Its final-status-response.json holds the CORRECT result and flag. 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-05-backup-spill single_solve mobile_puzzle /tmp/d05
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 5: Hatch Careful
Lane: Script analysis. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
script-analysis-day-05-python-carrier-lab - Prompt: "Nine Python behaviors come wrapped in fourteen disguises: bytecode, wheels, zipapps, frozen executables, and armored runtimes. Peel each carrier back to its intent without letting anything hatch, and identify the package that tells a different story than its payload."
What we were given
- One
evidence.zip(49,127 bytes, SHA-2560ad1ebe9953e7dde6d4ecdc54d7e4309d7547c9c7cb2bc842f51f4156169a599). Password:infected. - The Day 4 set (PowerShell, JScript, Batch, VBScript) plus 14 new Python files.
- The Python files are already-recovered source, one per "carrier". There were no real
.exe,.pycor wheel files to unpack. Each file name says which carrier it stands for (pyinstaller-onefile-recovered.py,wheel-recovered.py,zipapp-recovered.py, ...). - Each Python file hides its config in two constants, like VBScript on Day 4:
MATERIAL_<hex>: a list of numbers, a decimal string, ab"\x.."byte string, Base64 chunks in brackets, or URL-safe Base64.CODEC_<hex>:representation|transform|key. Real examples:base64_chunks|reverse_then_xor|29,reversed_integer_list|add_modulo_256|18,urlsafe_base64|add_modulo_256|05.
What was asked
- 357 questions: the 231 from Day 4 plus 126 Python questions (9 per file).
- Python questions per file:
- Role and ATT&CK ID from the decoded behavior graph (the question text gives the rules).
- Carrier (for example
pyinstaller_onefile), Windows chain (for examplemshta_to_python), syntax and obfuscation profile. - Adapter, campaign ID, and one indicator (URL, file path, next stage, host ID, upload URL, handoff ID or extension).
- Each question's
artifactfield names the exact file, for exampleevidence.zip -> zipapp-recovered.py. So there was no guessing which file a question meant.
How it was solved, start to finish
- Started the untimed Single Solve and downloaded the archive.
- Reused
solve_expansion.pyfor the 231 Day 4 questions. - Wrote a Python decoder (
pydec.py):
- It reads the
MATERIAL_value with Python'sastmodule (ast.parse+ast.literal_eval). That works for every literal form (lists, tuples, byte strings, joined strings) and never runs the file. - It then reuses the Day 4 transform search. The text must match
v=<n>|<key>=.
for node in ast.walk(ast.parse(source)):
if isinstance(node, ast.Assign):
for t in node.targets:
if isinstance(t, ast.Name) and t.id.startswith('MATERIAL_'):
material = ast.literal_eval(node.value) # data only, nothing runs
Example: pyinstaller-onefile-recovered.py is hex_colon + single_byte_xor and decodes to:
v=2|steps=acquire_remote,write_file|adapter=http_client_to_path|stage_count=1|syntax=generator_pipeline|obfuscation=fstring_fragments|carrier=pyinstaller_onefile|windows_chain=mshta_to_python|campaign_id=campaign-60cc9d6d2f|url=https://60cc9d6d2f.example.invalid/93aeb91b.py|user_agent=HuntressPython
- Carrier, Windows chain, syntax and obfuscation are fields in the decoded config, so they were copied as-is.
- Role from steps. Python's Command and Scripting Interpreter ID is
T1059.006. One URL-safe Base64 file failed at first, and the decoder was fixed for it. - Checked with the partial-submit oracle (see Day 3). Only
python:role(14 answers) was rejected. Trying each role value showed the cause: the two relay files wantrelay, notstaged_chain. After that fix the Single Solve wasCORRECT.
| File | Carrier | Role |
|---|---|---|
source-context.py | raw_source | benign_utility |
pyarmor-source-recovered.py | pyarmor_obfuscated | stager |
wheel-recovered.py | wheel_archive | stager |
pyinstaller-onefile-recovered.py | pyinstaller_onefile | downloader |
pyinstaller-onedir-recovered.py | pyinstaller_onedir | uploader |
pyc38-recovered.py | hash_based_pyc_38 | dropper |
cxfreeze-msi-recovered.py | cx_freeze_msi | dropper |
cxfreeze-recovered.py | cx_freeze_executable | impact |
current-pyc-recovered.py | hash_based_pyc_current | launcher |
py2exe-console-recovered.py | py2exe_console | launcher |
py2exe-service-recovered.py | py2exe_windows_service | staged_chain |
pyarmor-pyinstaller-recovered.py | pyarmor_pyinstaller | staged_chain |
embedded-runtime-recovered.py | embeddable_python_runtime | relay |
zipapp-recovered.py | zipapp_pyz | relay |
- Time Trial and Mastery first failed inside the driver, for two reasons:
- one variant stored
MATERIAL_as a tuple of strings, which needed joining; run_expansion.pydid not pass thequestionslist to the solver, and the Python part needs it to map IDs to files.
After both fixes, the Time Trial (50-second deadline) and Mastery (714 questions in two sets, 80-second deadline) both came back CORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 4c0dc263e3563c73cb37d6fe2a38b002 |
| Time Trial | CORRECT | ff6fe6d53dd879aee6c5ba0e19701844 |
| Mastery | CORRECT | 320bf533e8a436d37652ee605c580d24 |
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/d05/ex
7z x -pinfected -o/tmp/d05/ex repos/day-05-script-analysis-hatch-careful/single-solve/evidence.zip
python3 -I tools/solve_expansion.py /tmp/d05/ex \
repos/day-05-script-analysis-hatch-careful/single-solve/start-response.json /tmp/d05/answers.json
This rebuilds all 357 saved answers (re-checked on 2026-11-01). It only parses the files.
Live: python3 -I ~/.config/huntress/solvers/run_expansion.py script-analysis-day-05-python-carrier-lab <mode> <workdir>. This needs a Game Master token. The solvers/ copy runs JScript fixtures in a Node vm sandbox; the tools/ copy does not.
Day 5: By the Numbers
Lane: Warmups. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
warmups-day-05 - Prompt: "When you think about it, every letter is just a number."
- New encoding: decimal ASCII
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. Space-separated decimal numbers. Each number is an ASCII code point; convert to its character.
- Mastery blobs use multiple encodings from Days 1-5. 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:sturdy-river-grove-57. Flag:8c7470078c17018b63e67d0d1aeb3284. - Time Trial:
CORRECT. slug:kind-cloud-willow-54. Flag:8bc8d53ca90fb3ea55ad1ac5d29d540c. - Mastery:
CORRECT. 23 slugs. Flag:294b5275ceb2110b6c2a206908b85931.