Tactical LogHuntress CTF 2026 // team archive

Days / Day 29 / Warmups / Rotten Base

DAY 29//LANE: WARMUPS//STATUS: CORRECT x3

Rotten Base

Warmups SSTTMM Modes reported CORRECT in this write-up

Day 29 - Rotten Base (warmups)

Prompt: "Sixty-four invited thirteen over. Things got scrambled." -> base64 (sixty-four) with ROT13 (thirteen) applied over it.

Results (all modes CORRECT)

ModeResultFlag
Single SolveCORRECT (3.2 s)Token02a9e14bde2f3557a64c8599f405db9c
Time TrialCORRECT (5.1 s, ranked)Token13277247ba71107c66a9f2835005a3f5
MasteryCORRECT (16.3 s, regular retry path)Token3deda2e831cf6a68e6449f86c8793992

Single / Time Trial

One blob. Peel ROT13, then base64, to reach the carrier sentence "Nice work decoding this warmup! To solve, submit this exact answer to the Game Master: <slug>". The answer is the word-word-word-NN slug. decode_chain.py finds rot13 > base64 directly; the existing warmups adapter already handled it.

Mastery

143 blobs (blob-01.txt .. blob-143.txt, ids set1.message .. set143.message), each a 29-row numbered list, one encoding per row, whose decoded rows concatenate to the carrier sentence plus the slug. Rows 1-28 are the Day 28 ladder (base64, rot13, hex, binary, decimal, octal, percent, html, json-\u, base32, base64url, reverse, atbash, caesar+7, rot47, morse-hex, bacon-hex, literal, utf16 hex x2, gzip-b64, zlib-b64, ascii85, rle "N:C,", U+ codepoints, brainfuck, Ook!, reversed hex). Row 29 is NEW: ROT13 over base64 (e.g. oP0kAj== -> rot13 bC0xNw== -> l-17). It sits in the unconstrained slug tail, so the known-sentence guide cannot pick it and the adapter correctly refused to guess.

What went wrong first, and the fix

The first ranked Mastery attempt raised mastery blob for set1.message did not validate against the known sentence (not submitted). That was the NO-GIBBERISH guard working: the solver had no row-29 decoder, so the slug tail was unreadable. Nothing was submitted and no variant was spent. Fix (private clone scripts/warmups_wip.py only): added _rot13_b64 and LINE_DECODERS["rot13b64"], plus LADDER_NAMES_D29 = LADDER_NAMES_D28 + ["rot13b64"] selected when a blob has 29 rows. The validator was left strict (relaxing it would have risked submitting a truncated slug). Offline check on the 143 downloaded blobs: 143/143 decode with the full known sentence; set1 -> jolly-stream-hazel-17; Single unchanged. The ranked attempt had already ended, so the runner continued with a regular retry on a fresh variant and Mastery was accepted.

Lessons

  • A slug-tail row beyond the known ladder needs a ladder entry; do not relax the known-sentence validator.
  • Daily layer theme (title/prompt) predicts the new row: "Sixty-four ... thirteen over" = rot13 over base64.