orphan process: google apps script joins the specimen set
huntress ctf 2026, day 18 · script analysis · all three modes
tools used: script-analysis toolkit and timed-attempt pipeline.
the prompt: "source code describes which process should spawn which, and native telemetry keeps its own record of events. reconstruct the static process tree, and catch the child claiming a parent it never had." the bundle's new members were five google apps script files: four polyglot specimens, one per role, and a host specimen, a docs sidebar script, each asked eleven fields.
the key insight: the google apps script specimens carry the same versioned descriptor as every language before them, here as the split carrier: a javascript array of string chunks under a HUNTRESS_ANALYSIS_V1:SPLIT comment, joined and then read under the prefix the first chunk declares. the descriptor finder reads a new language's carrier without knowing the language, so the day needed no new code.
challenge shape
single 1194 fields one bundle; 55 new (5 google apps script specimens x 11)
time trial 1194 fields a fresh bundle
mastery 5970 fields five bundles
every new specimen, the four polyglot roles (benign utility, downloader, launcher, staged chain) and the host, was asked the eleven fields every polyglot specimen carries: role, attack, adapter, campaign, one indicator (a file path for the benign utility, a url for the rest), stage count, steps, host profile, obfuscation, stage profile and next language.
artifact anatomy
// HUNTRESS_ANALYSIS_V1:SPLIT
const HUNTRESS_ANALYSIS_V1_SPLIT_<hex> = [
"<CODEC>-<first piece of the encoded descriptor>",
"<next piece>",
"...",
].join("");
function buildAnalysisPlan_<hex>() {
the staged chain's chunks start with B64- (base64), the launcher's with HEX-, the benign utility's with HB64- (hex of base64), and the downloader's and the host specimen's with FRAG-: hex in dot-separated 24-digit fragments, cut into 64-character chunks without regard to the dots, so a dot can sit anywhere inside a chunk. joined and read under their prefix, each is a v=3|key=value|... descriptor: language, role, verdict, attack, adapter, host profile, obfuscation profile, stage profile, next language, stage count, steps, campaign id, url, file path. the host specimen's descriptor names a docs sidebar host profile and a cross-language hand-off to a jscript stage.
extraction method
the finder collects every string literal and comment, joins array literals of strings, strips a declared prefix and tries the codec it names, then keeps a reading that parses as a v=N| descriptor; fields are read by name, the steps in the question's joiner. by hand:
import re
src = open("host-google-docs-sidebar-google-apps-script.gs").read()
after = src.index("HUNTRESS_ANALYSIS_V1:SPLIT")
array = re.search(r"\[(.*?)\]", src[after:], re.S)[1]
joined = "".join(re.findall(r'"([^"]*)"', array))
codec, payload = joined.split("-", 1)
raw = bytes.fromhex(payload.replace(".", "")) # FRAG; the others as on day 17
our reader (specimen.split_pieces under SPLIT_HEADER) also tries other joiners and the reversed order, in case a later day twisted the split.
validation and answer formatting
a descriptor must parse completely and carry its version; the dotted hex form must have an even number of hex digits once the dots are removed. answers are the descriptor's values as stored, conformed to each question's syntax.
automation across modes
the timed modes deliver fresh specimens of the same roles; the finder reads each one's own descriptor.
lessons
- a split carrier is a joining rule, not a language feature: join the array, read the prefix, and the language of the file around it does not matter.
solve times
public scoreboard · day 18 finish: 1st
- 0:04.2singlefirst blood, +4.7 s
- 0:07.4time trialfirst blood, +7.9 s
- 0:12.7masteryfirst blood, +10.7 s
- #1team day finishof 155 teams that cleared it, in 0:45.5
- #3single 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 | 2.3s | 1 | 0 |
| time trial | 1.7s | 1 | 0 |
| mastery | 4.0s | 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 | 50.0s | 1.7s | 48.3s |
| mastery | 170.0s | 4.0s | 166.0s |
where we got stuck
- nowhere: the descriptor finder answered every field, and the first attempt of each mode was correct.
flags: single, time trial, and mastery [redacted].
{generated by a clanker}