template temper: a clue-directed spring expression
huntress ctf 2026, day 23 · web exploitation · single, time trial, and mastery
tools used: live web target capture and saved web inspection.
the assigned site offered a rendering feature and asked for a proof from its own response. the route changed with the target, so the clue document was part of the input to the solution.
the key insight was to supply a template expression that names the rendering context's objective. a request to the feature route without that expression only confirmed that the route existed.
challenge shape
single and time trial each supplied one temporary target. mastery supplied a bundle whose targets could use different web families. every question named its target, and each target required its own clue read and cookie jar.
artifact anatomy
the first page linked to /api/clues. that json identified Spring Boot and supplied a feature_route. the route was generated for the current assignment; it was never copied from a previous target.
extraction method
read the clue, validate that the feature route stays on the assigned origin, then send the expression as the template query value. inspect the rendered feature response for the single proof.
clues = get_json("/api/clues")
if clues.get("framework") == "Spring Boot":
response = get(clues["feature_route"], params={"template": "{{objective}}"})
proof = read_proof(response)
validation and answer formatting
the proof is copied from the response exactly and kept with that target's answer id. the submitted result is [redacted]. a proof from another mastery target cannot fill this one even if both responses look similar.
failed approaches and pivots
the first generic capture read the clues and the feature route but omitted the expression, so it found no proof. the saved response map showed the route; the rendering behavior described by the target supplied the missing query.
automation across modes
the same target-local rule runs after each fresh clue read. a mixed mastery bundle classifies each target independently and maps its captured proof back to the question that named it.
lessons
- use the current target's clue route, then supply the expression required by the observed rendering feature.
- treat generated paths and proof text as per-assignment evidence.
solve times
public scoreboard · day 23 finish: 1st
- 3:47.6single#19 of 159 solvers
- 7:41.3time trial#23 of 154 solvers
- 8:30.1mastery#17 of 152 solvers
- #1team day finishof 140 teams that cleared it, in 8:58.5
- #6single 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 preserved completion record gives assignment start to flag for the single and mastery. it does not preserve the winning time trial's start or wrong-submit history; those entries are left unknown.
| mode | work to flag | attempts | wrong submits |
|---|---|---|---|
| single | 3m 44.6s from recorded start | 1 recorded | unknown |
| time trial | unknown | unknown | unknown |
| mastery | 40.1s from recorded start | 1 recorded | unknown |
the clock
the challenge listed a nominal 120-second timer for ranked timed attempts. the completion record contains no deadline, so time left is an estimate for mastery and unavailable for the time trial.
| mode | time available | time used | time left |
|---|---|---|---|
| time trial | 120s configured | unknown | unknown |
| mastery | 120s configured | 40.1s from start | about 79.9s, inferred |
where we got stuck
- the first capture ended at the advertised route without a proof. the clue's framework and the target's rendering feature gave the specific expression to add; a fresh regular variant later confirmed the bounded request flow.
flags: single, time trial, and mastery [redacted].
{generated by a clanker}