Tactical LogHuntress CTF 2026 // team archive

Days / Day 08 / Warmups / Entity Relations

DAY 08//LANE: WARMUPS//STATUS: CORRECT x3

Entity Relations

Warmups SSTTMM Modes reported CORRECT in this write-up

Challenge

  • Category/lane: Warmups (Huntress CTF 2026, Day 08)
  • GM id: warmups-day-08
  • Prompt: "The browser understands. Do you?"
  • Modes: Single Solve, Time Trial, Mastery (Mastery = 38 new dynamic tasks, per the GM's own Mastery prompt text).
  • Answer format: a single question id message; the answer is the exact slug a fully-decoded blob's final non-empty line instructs - "three lowercase words separated by hyphens, followed by a hyphen and a two-digit number; submit the slug only" (trim normalization).

Triage

Fingerprinted instantly against the warmup_decode family in challenges/_lib/kb/families.json (not known-hard, Sonnet-medium base tier - borne out: this challenge solved in seconds). The title itself is the hint, consistent with the family's naming-pun convention from every prior day ("Half a Turn" = ROT13, "Eight Is Enough" = octal, "Percent Chance" = percent/URL encoding). "The browser understands" points at a browser-native encoding, not a general-purpose one.

Investigation

Ran the assigned blob through challenges/_lib/decode_chain.py (the universal layered decoder already banked for this family); it fully decoded in one automated pass with no manual layer identification needed, and the final decoded line gave the exact answer slug per the GM's stated format.

Flag Recovery

Solved via challenges/_lib/autosolve_warmups.py, the day-generic adapter for this family - it ran automatically at claim time, before any agent reasoning was needed. All three modes graded CORRECT independently; Mastery's 38 fresh dynamic tasks (not a repeat of Single Solve) were each derived from that attempt's own assigned bytes, never a memorized value.

Flag

  • single_solve 3 seconds after release - Tokena43c961cb1da97df79b39acf66e3196a
  • time_trial Token914aa968abf54fd2de978b26f8d1ecca
  • mastery Tokencd267453f538104082d41325e5d32098

All three modes were solved and graded CORRECT by 2026-10-08T13:00:07Z, 7 seconds after the 09:00 ET release - this was the day's First Blood.

Minimal Reproduction

  1. Fingerprint the challenge against warmup_decode (title pun names the encoding layer).
  2. Run challenges/_lib/autosolve_warmups.py against the challenge id; it downloads the assigned blob, runs decode_chain.py's layered decoder, and submits the final line's slug.
  3. Mastery reuses the identical adapter against each of its 38 freshly-assigned tasks in one run.

Lessons Learned

  • The fastest lane in the game is also the one most completely automated end-to-end (claim -> autosolve -> submit, zero model reasoning) - this is the target shape for every other lane's adapter as it matures.
  • Warmups still gets a full writeup like every other challenge - "trivial" does not mean "undocumented" (corrected 2026-10-08; a prior convention of skipping the warmup lane's writeup was wrong and is retired starting this entry).