2026-10-01
Day 1: ELF Aware
Lane: Binary reverse engineering. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
binary-reverse-engineering-day-01-c-elf-orientation - Prompt: "Before you can reverse anything, you have to learn to read. These plain C binaries are your primer: headers, section tables, symbol tables. Start with
readelf -h -Sand work your way down."
What we were given
- Single Solve: one
evidence.zip(1,905 bytes, SHA-25690bb41ce9d55b924436486cb123db1d0a79a885697599511c44da5c336fad9b4). - It holds one 64-bit Linux program:
symbol-orientation.elf. - Day 1 zips have no password. A plain
unzipworks. - Time Trial: the first attempt gave one file,
section-beacon.elf. The retry gave four files,slot-0.elftoslot-3.elf(retry-evidence.zip, SHA-25648850ef76d65038de8f431ce86dffbb67524fb4894d5da4640050b17117b0ab7). - Mastery: two deliveries. The first try had one file per delivery. The retry had four files per delivery (
m-slot-0.elftom-slot-3.elf).
What was asked
- 9 questions per file, all about the ELF file itself:
- Single Solve: 9 questions (1 file).
- Time Trial retry: 36 questions (4 files).
- Mastery retry: 72 questions (2 deliveries × 4 files).
- Question types:
- Header facts: ELF class, byte order, CPU architecture, object type, and entry point (lowercase hex with
0x). - Sections: the text in the custom
.huntresssection, and whether a.symtabsection exists. - Symbols and code: the designated analysis symbol, and the constant that
recover_codereturns.
How it was solved, start to finish
Day 1 was solved by hand with readelf and objdump. These tools only read the file.
- Started the attempt and downloaded the zip. Unzipped it with
unzip -oq evidence.zip -d files. - Read the header and section table with
readelf -h -S -sW:
- ELF64, little endian,
EXEC, x86-64, entry point0x401030. - The section list includes
.huntressand.symtab.
- Read the marker with
readelf -x .huntress. It holdsBR01-SYMBOL-ORIENTATION. - Found the returned constant.
objdump -d -M intel --disassemble=recover_codeshowsmov eax,0x539thenret.0x539is 1337.
Re-checked with a small Python parser (symbol table → function bytes):
recover_code: vaddr=0x401120 size=6 file_off=0x1120 bytes=b8 39 05 00 00 c3
mov eax, 0x539 -> 1337
- Submitted the 9 answers in
answers.json. Examples:"elf_type": "ET_EXEC","marker": "BR01-SYMBOL-ORIENTATION","recover_code": "1337","entry_point": "0x401030". Result:CORRECT. - Time Trial:
- The first timed attempt expired. Its submit came back
incorrect. - A retry gave four files. Three were the same program. The fourth (
slot-3.elf) had markerBR01-IMMEDIATE-RECOVERYand returned 7331. - All 36 answers were
CORRECT. Per-file values are incommands-output.txt.
- Mastery:
- The timed attempt expired with two fields unconfirmed (submit result
incorrect). - The Game Master then allowed untimed practice (
op=practice) on the same two retry deliveries. - Some files had no
recover_codesymbol. Their designated symbol wasinspect_sectionand their marker wasBR01-SECTION-BEACON. - A partial practice submit accepted 70 of 72 answers. The two missing fields were
set1.slot3_recover_codeandset2.slot1_recover_code, the two files withoutrecover_code. - The Game Master accepted
nonefor those two fields. The full set of 72 answers was thenCORRECT.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | f61edda783482ec70bf509132cc25d41 |
| Time Trial | CORRECT | 16883e2891052e3b4cfcb0e25384560c |
| Mastery | CORRECT (practice) | 3d6d6d0d1cd19d185c436ee82db7c54f |
Each mode folder has the zips, the extracted ELF files, the answers, the command output, and the raw Game Master responses. No screenshots were taken.
Reproduce
cd single-solve
unzip -oq evidence.zip -d files
readelf -h -S -sW files/symbol-orientation.elf
readelf -x .huntress files/symbol-orientation.elf
objdump -d -M intel --disassemble=recover_code files/symbol-orientation.elf
These commands only read the file. They never run it.
Day 1: You Must Be This Hash to Ride
Lane: Metadata forensics. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
metadata-forensics-day-01-hash-to-ride - Prompt: "Three ordinary files arrived without trustworthy labels. Build an identity inventory from the bytes themselves and prove which labels can—and cannot—be trusted."
What we were given
- Single Solve:
(372 bytes, SHA-256metadata-day-01-evidence.zipa511416df9e58c2a85f93a35a3c342c6cf8e03abc10d4b8eef1483394a5b64e5). It holds one file:(240 bytes).evidence-note.txt - Time Trial:
(910 bytes, SHA-256metadata-day-01-evidence.zipf12d4fe9d91748f95b6d7fbf3b3bae7012b8b97e4200e6f9d60f0bba92a79920). It holds three files:evidence-note.txt(240 bytes),evidence-note-copy.txt(240 bytes) androster.csv(129 bytes). They are saved in.time-trial/files/ - Mastery:
(2,122 bytes, SHA-256mastery-bundle.zip2d40947bd7030416df7b620f377c38ba85c3e19b5fdfd35d9225c26f27fce937). It holds two inner zips, one per set:metadata-day-01-evidence-01.zipandmetadata-day-01-evidence-02.zip. Each has the same three file names. The inner zips and their files are in.mastery/files/
What was asked
- Single Solve: 5 questions about the first evidence note: exact filename, size in bytes, MD5, SHA-1 and SHA-256.
- Time Trial: 5 questions spread over
roster.csvand the two notes. - Mastery: 10 questions, 5 per set. Question IDs start with
set1.orset2.. - Question types:
- Filename: the exact base name, for example "the renamed evidence-note copy".
- Size: a base-10 byte count.
- Hashes: MD5 (32 hex), SHA-1 (40 hex), SHA-256 (64 hex). Letter case is ignored.
How it was solved, start to finish
These were solved by hand with standard command-line tools. Each mode folder has a commands-output.md with the exact commands and output.
- Started the attempt. The reply listed the questions. Each question names its file, for example
metadata-day-01-evidence.zip -> evidence-note.txt. - Downloaded the zip and listed it with
unzip -lto get the exact member names. - Extracted the files. The mastery inner zips were extracted with
7z x -pinfected. - Measured and hashed each named file:
wc -c files/evidence-note.txt
md5sum files/evidence-note.txt
sha1sum files/evidence-note.txt
sha256sum files/evidence-note.txt
Single Solve output:
240 files/evidence-note.txt
6b5074cd2b741ee5d0f0f9cae5eba7d5 files/evidence-note.txt
11f594f647b3c81bce2527c0d9f6c72a60f492d9 files/evidence-note.txt
a6d2d239a6ba996a16ea585c9dd5874b9a0fcdb1e16e3f10457d5ea79e944dc6 files/evidence-note.txt
- Checked which labels can be trusted. In the Time Trial zip and in both mastery sets,
evidence-note.txtandevidence-note-copy.txthave the same bytes. So the "copy" label is true.roster.csvis a different file with its own hashes. - Submitted every answer under the exact answer IDs from the start reply. Examples from the answer files:
- Time Trial:
roster-filename=roster.csv,roster-md5=47ac140f16634f1a2e64a44224756592. - Mastery:
set1.roster-sha256=773bdc1c478256227830fce107daf399542c7668aa80fabf1560bc3a3a4f5eb6,set2.evidence-note-size-bytes=240.
- The Game Master replied
CORRECTfor each mode and returned the flag.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 4079fe133b7df492ee6f974989caec22 |
| Time Trial | CORRECT | 3292cc8f4328d673a49f7d048290e048 |
| Mastery | CORRECT | 3273f32b95bf615f1a0c75ca4630de18 |
Each mode folder has start-response.json, answers.json, submit-response.json, commands-output.md, the original zip and the extracted files. No screenshots were captured.
Reproduce
cd single-solve
unzip -l metadata-day-01-evidence.zip
wc -c files/evidence-note.txt
md5sum files/evidence-note.txt; sha1sum files/evidence-note.txt; sha256sum files/evidence-note.txt
The same field types are handled later by ~/.config/huntress/solvers/run_meta.py (-filename, -size-bytes, -md5, -sha1, -sha256). Nothing here runs the files; it only reads their bytes.
Day 1: Exported Trust
Lane: Mobile (Android). Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
mobile-android-day-01-exported-trust - Prompt: "Identify an exported deep-link Activity that attaches an embedded client credential, invoke it with the correct case reference, and return the backend evidence code."
What we were given
- One Android APK per case, 12,704 bytes each. Inside:
AndroidManifest.xml,classes.dex,assets/huntress_puzzle.json, and signing files inMETA-INF/. - Single Solve:
exported_trust.apk, caseMOB-01-073E9A824E, SHA-25664169ad2f314746e4d82cd048cd3c436dffafdedc187e01f44cabe6c0068509f. - Time Trial:
exported_trust_time_trial.apk, caseMOB-01-ED1C7717EE, SHA-2569ddef61a7a5ec9a1e65765fe11163af613b886f9cf782b0811ea9c6169119a2d. - Mastery: three APKs, one per case (
MOB-01-0DAC1DD945,MOB-01-D28E90FE08,MOB-01-F1EC220721). Their SHA-256 values are inmastery/commands-output.txt.
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. Everything came from the APK bytes.
- Started the attempt and downloaded the assigned APK from the Game Master.
- Read the manifest with
aapt dump xmltree. It shows an exported.ChallengeActivitywith aVIEW+BROWSABLEfilter forhuntress01://exported-trust/unlock. - Read the DEX.
ChallengeActivityreads theanswerandcase_idvalues and forwards them to the evidence endpoint. - Tried the local backend. The request from the VM got "connection refused". So the code was taken from the answer bundled in the APK instead.
- Read the puzzle asset. The codec is
plain, sodatais already the answer:
unzip -p exported_trust.apk assets/huntress_puzzle.json
{"schema":"huntress.mobile.puzzle.v1","day":1,"codec":"plain","data":"ANS-DACC86F43B0B1C1A"}
- Submitted
ANS-DACC86F43B0B1C1Aunderanswer. The Game Master repliedCORRECT. - Time Trial: downloaded the APK with
op=artifact, read the same asset. It gaveANS-2010BA5274DFFAE1. - Mastery: read the asset in each of the three APKs. The codes were
ANS-4324FFECDA397A2A,ANS-946ED16E5EC11EF6andANS-E4B93CC458BAFB45. All three went in one submit.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 15da3bfcc83a96a7c1bc7f8a7ff79e3d |
| Time Trial | CORRECT | d041da5da21c72e69ae6052384192aaf |
| Mastery | CORRECT | 180a5d968c1d7031f88533a715b5404b |
Each mode folder has the APK(s), the extracted puzzle, answers.json, start-response.json, submit-response.json, and a commands-output file with the commands and output. No screenshots were taken.
Reproduce
Offline, from the saved APK:
unzip -p single-solve/exported_trust.apk assets/huntress_puzzle.json
A fresh run with the later lane driver (it handles the plain codec):
python3 -I ~/.config/huntress/solvers/run_mode.py mobile-android-day-01-exported-trust single_solve mobile_puzzle /tmp/d01
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 1 — Read the Rules
Lane: (introductory) · Challenge ID: read-the-rules
> Read the rules page. Find the hidden flag.
What we were given
A link to the CTF rules page at https://ctf.huntress.com/rules.
How it was solved
- Downloaded the rules page. It's a Next.js app — the static HTML shell loads a route script.
- The route script uses
dangerouslySetInnerHTMLto inject an HTML comment into the page. The comment contains the flag. - Viewing page source (or inspecting the JS bundle) reveals the flag string.
- Submitted the flag to the Game Master.
This is a warmup — the lesson is "always view source."
Results
| Mode | Result |
|---|---|
| Single Solve | CORRECT |
Day 1: Power Play
Lane: Script analysis. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
script-analysis-day-01-native-signals - Prompt: "Three PowerShell specimens are on the table, and none of them will ever run. That's fine, because you only need to read them. Classify each script's intent, map its primary ATT&CK technique, and pull out the exact indicators sitting in plain sight."
What we were given
- One
evidence.zip(2,465 bytes, SHA-25607e61f33b72ae646d925087de9b9e94068812fdf12f7e43613d0b2bef6f6d3e5). It is AES-encrypted with the passwordinfected. Extract it with7z x -pinfected evidence.zip. - Three PowerShell files:
resource-sample.ps1,collection-sample.ps1andimpact-sample.ps1. - Each file holds one function. Inside it is a call to
[Convert]::FromBase64String('...'). The Base64 text is a small JSON "descriptor" with one-letter keys (a,b,c, ...).
What was asked
- 15 questions, 5 per file. Each question names its file.
- Question types:
- Scenario: one of
resource_hijacking,synthetic_collection,synthetic_impact. - ATT&CK ID: the primary technique. For
collection-sample.ps1the question said to treat browser credential-store access as the main behavior. It also accepted the broader "Data from Local System" ID. - Exact indicators:
- Resource sample: URL, wallet, mutex.
- Collection sample: URL, first file name, user agent.
- Impact sample: file extension, recovery-note file name, URL.
How it was solved, start to finish
- Started an attempt with the Game Master (
op=start). It returned the prompt and the 15 questions as JSON. - Downloaded the artifact and extracted it with 7z. Three files came out.
- Read the files as text. Nothing was run.
- Decoded each descriptor. A regex finds the Base64 string, then it is decoded and parsed as JSON. This is the exact command from
single-solve/commands-output.md:
import base64, json, pathlib, re
for path in pathlib.Path(".").glob("*-sample.ps1"):
encoded = re.search(r"FromBase64String\('([^']+)'\)", path.read_text()).group(1)
print(path.name, json.loads(base64.b64decode(encoded)))
Output (Single Solve):
resource-sample.ps1: {"a":"a88bfda72dbe","b":"https://cdn-03ce326948.example.invalid/3e59094b24e2.bin","c":"wallet_d327bc1114a907589af81bba71226b77","d":"update-03ce326948.dat","e":"HKCU\\Software\\HuntressLab\\Run\\03ce326948","f":"Mozilla/5.0 HuntressLab/3.6","g":64,"h":"Global\\HuntressLab_6460b4f6aa72"}
collection-sample.ps1: {"a":"e536c782dd01","b":"https://497b6b9b43.example.invalid/3f890532dd42.dat","c":["update-497b6b9b43.dat","Login Data","Cookies"],"d":"stage-e536c782dd01.zip","e":"Global\\HuntressLab_03b15c606290","f":"Mozilla/5.0 HuntressLab/3.3"}
impact-sample.ps1: {"a":"785244c7fdf5","b":".locked-1d685","c":"RECOVERY-1d6858b777.txt","d":"https://1d6858b777.example.invalid/a0e20204ed17.dat","e":["update-1d6858b777.dat","Login Data","Cookies"],"f":"Global\\HuntressLab_b2e257a4ea0f"}
- Picked each indicator by its shape. The key letters move around between files, so the value's form is what matters:
- URL: starts with
http://orhttps://. - Wallet: starts with
wallet_. Mutex: starts withGlobal\. User agent: starts withMozilla/. - Extension: starts with
.. Recovery note:RECOVERY-*.txt. - First file name: the first item of the list (
update-*.dat).
- Mapped scenario and ATT&CK by file. A few real answers from
single-solve/answers.json:
{"resource-scenario": "resource_hijacking", "resource-attack": "T1496",
"collection-scenario": "synthetic_collection", "collection-attack": "T1555.003",
"collection-file": "update-497b6b9b43.dat", "collection-user-agent": "Mozilla/5.0 HuntressLab/3.3",
"impact-scenario": "synthetic_impact", "impact-attack": "T1486", "impact-note": "RECOVERY-1d6858b777.txt"}
- Submitted all 15 answers as one JSON object. The Game Master replied
CORRECTand returned the flag. - Time Trial: a new
evidence.zip(SHA-25676980d6a5ad47ff42da017569278ceaf77bc9f226090bd5fa5d947c3caf5480e) with fresh values. The deadline was 50 seconds after the start. Same steps. The decoded values are intime-trial/commands-output.txt. - Mastery: 30 questions in two sets (
set1.*,set2.*), with an 80-second deadline. The download wasmastery.zip(5,166 bytes, SHA-256f96ad10b598a63dc897fd8890dd7de8cbbd9cbc1717d3f72627484d057b1b2cd). It holdsevidence-01.zipandevidence-02.zip, one per set. Each set was decoded the same way, and the answers were sent under theirsetN.IDs. Seemastery/commands-output.txt.
Results
| Mode | Result | Flag |
|---|---|---|
| Single Solve | CORRECT | 8011b2699059fed172501b192bc44e2b |
| Time Trial | CORRECT | d89beb994e05ea968d54e417c93b8bed |
| Mastery | CORRECT | 341deea112be2d8eb47db3825fe2ff98 |
Each mode folder keeps the archive, the extracted .ps1 files, start-response.json, answers.json and submit-response.json. No screenshots were taken.
Reproduce
Offline, from the repo root, using the saved Single Solve files:
mkdir -p /tmp/d01/ex
7z x -pinfected -o/tmp/d01/ex repos/day-01-script-analysis-power-play/single-solve/evidence.zip
python3 -I tools/solve_expansion.py /tmp/d01/ex \
repos/day-01-script-analysis-power-play/single-solve/start-response.json /tmp/d01/answers.json
This rebuilds all 15 saved answers (re-checked on 2026-11-01). The solver only reads the .ps1 files as text.
Day 1: A Little Based
Lane: Warmups. Modes: Single Solve, Time Trial, Mastery. Result: all CORRECT.
The challenge
- ID:
warmups-day-01 - Prompt: "Everyone has to start somewhere. I hear sixty-four is a good base."
- New encoding: base64
What we were given
- One blob file per mode (Single Solve and Time Trial), or a zip of multiple blobs (Mastery).
- The blob is a single encoded string.
- The decoded text ends with a slug like
word-word-word-NNto submit.
How it was solved
- Decode the blob. Base64-decode the blob to get plaintext.
- The decoded text contains a message ending in the answer slug.
- Submit the slug to the Game Master.
Results
- Single Solve:
CORRECT. - Time Trial:
CORRECT. - Mastery:
CORRECT.
Day 1 — Welcome Robots
Lane: (introductory) · Challenge ID: welcome-robots
> Ask your AI teammate to identify the Welcome Robots challenge and provide the answer.
What we were given
A prompt asking for the AI teammate's answer.
How it was solved
- Queried the Game Master API for the Welcome Robots challenge.
- The Game Master returned the flag directly — this was an introductory challenge testing that the AI Arena tooling was working.
- Submitted the returned flag.
This is a connectivity/setup check, not a real challenge.
Results
| Mode | Result |
|---|---|
| Single Solve | CORRECT |