shell company: following a path into a shell

huntress ctf 2026, day 25 · web exploitation · three modes

tools used: live web target capture, saved web inspection, and timed-attempt pipeline.

the target looked like a small workspace application. its clue response named a feature route, a path parameter and an example path. the feature passed that parameter through a command boundary, so a request meant to inspect a directory could also execute a second command.

challenge shape

single and time trial each assigned one temporary target and one answer. mastery assigned four targets and four answer ids. each assignment rotated the host and route. a mastery target could belong to another web family, so the title alone did not identify every target's method.

artifact anatomy

the useful artifact was the saved sequence of http responses. the home page linked to a clue document; the clue named the current feature route, the input parameter and a benign example path. requesting the feature without the parameter produced a required-input error. with the example, the response included the worker's command output.

extraction method

we first followed the current target's clue route, then supplied its example path under the named parameter. a harmless identity command appended to that path changed the output, confirming that the input crossed a shell boundary. we then read the challenge's objective file through the same parameter and current route.

clues = read_current_target_clues()
route = clues["feature_route"]
key = clues["input_parameter"]
example = clues["example_path"]
baseline = get_same_origin(route, params={key: example})
boundary = get_same_origin(route, params={key: example + "; id"})
if command_output_changed(baseline, boundary):
    evidence = read_challenge_objective_through_current_route(route, key, example)

the objective path and each proof are omitted here. they are not reusable between assignments. the proof had to appear in this target's response before it was submitted for that target's answer id.

validation and answer formatting

the flow runs only when the current clues identify the expected framework and supply all three inputs. every request remains on the assigned origin. an http success status or a directory listing is evidence, not the answer; only the returned proof is used. mastery keeps each target's response paired with its own answer id.

failed approaches and pivots

the initial generic collector requested the feature without the required parameter and got an error. a later request with the example path returned only the directory listing. listing that directory and its parent confirmed command execution but did not locate the objective. the saved response map made the method and parameter clear; the local handoff supplied the bounded objective-file read that finished a separate blind test.

automation across modes

the same clue-derived route and parameter logic applies to fresh time-trial and mastery targets. it cannot carry the single's route or output forward. the shared collector now records the example request and one harmless command check in its first capture; a challenge-specific step still has to read the objective and verify the returned proof.

lessons

solve times

public scoreboard · day 25 finish: below top 10

  • 2:23.0single#25 of 156 solvers
  • 5:40.3time trial#43 of 152 solvers
  • 6:15.3mastery#29 of 149 solvers
  • #11team day finishof 138 teams that cleared it, in 6:15.3
  • #8single orderof the day's 8, by our flag time

from the public scoreboard: 09:00 et release to our flag; the table below counts from our attempt's start.

the table measures from the first local work to recorded team completion. the timed team completions did not retain a winning local capture; these figures are not winning-attempt runtimes.

mode work to team completion local attempts wrong submits
single about 140s 1 0
time trial about 193s 1 0
mastery not reliably measured not recorded 0

the clock

the winning timed attempts' assignment deadlines were not preserved in the local record.

mode time available time used time left
time trial not recorded not recorded not recorded
mastery not recorded not recorded not recorded

where we got stuck

flags: single, time trial, and mastery [redacted].

{generated by a clanker}