Writeup · Cyberweek Africa 2026 - National CTF (Finals)

The Lost Signal 1: The Telegraph Ledger

forensicseasy-mediumsolvedFlag: FLAG{REDACTED}

Points: 150
R1 (retained for Chapter 2): 8EC5C9EBF7C3


Overview

A fictional archive recovered coastal telegraph traffic from an unidentified relay experiment. The original transmission is gone, but copies survive at three stations whose local clocks disagree. Some records were damaged in transit, and some records carry perfectly valid checksums despite being untrustworthy. The message was retransmitted.

The task is to:

  1. Identify the historical telegraph line described by the cited primary-source catalogue.
  2. Recover the authentic frames for case ARC-26251F and reconstruct the message.
  3. Compute the relay token R1 per operator_notes.txt.
  4. Submit FLAG{LS01_<R1>}.

All traffic, dates, logs, and checksums are synthetic artefacts; the historical citations refer to a real documented coastal telegraph route.

Recon — identifying the line

historical_references.txt cites two genuine sources:

  1. Kenya National Archives (KNADS) — the Mombasa–Lamu line as the first telegraph line in East Africa.
  2. British Library WOMAT/AFR/BEA/209 — “Mombasa (Frere Town) to Malindi, & Malindi to Lamu, Survey of Telegraph Line”, 1890–1892.

So the surveyed coastal route is Mombasa / Frere Town → Malindi → Lamu.

station_register.pdf (decoded with pypdf) lists:

Code Station Numeric marker
MSA Mombasa / Frere Town 03
MLD Malindi 11
LMU Lamu 19
NBI Nairobi 05
KSM Kisumu 09
ZNZ Zanzibar 25

And three candidate line IDs:

Line ID Recorded stations
LN-17 MSA > MLD > LMU
LN-23 MSA > NBI > KSM
LN-41 MSA > ZNZ > LMU

Only LN-17 (MSA → MLD → LMU) matches the Mombasa → Malindi → Lamu coastal route. LN-23 and LN-41 are decoys, as are all ledger rows carrying other line IDs or case IDs.

Line seed

Per operator_notes.txt, the line seed is weighted by station order:

seed = sum((position from 1) * station marker) mod 32
     = (1*3 + 2*11 + 3*19) mod 32
     = 82 mod 32
     = 18

Sanity check against the notes’ known example: plain WIRE 1 at q=00 on the coastal route encodes to wire 67MPQQ with CHECK 024. Encoding with seed 18 reproduces this exactly, confirming the seed.

Clock calibration

clock_sync.csv logs each station’s local clock against a common reference UTC, so offset = local − reference and reference = local − offset:

Logger Reference UTC Local Offset
MSA 08:00:00 08:09:00 +9 min
MLD 08:00:00 07:56:00 −4 min
LMU 08:00:00 08:06:00 +6 min

A normal frame is first logged at the opening station (MSA) at reference 09:00:00, then at MLD (+20 s) and LMU (+40 s); the next frame begins 120 s later. So frame q is expected at:

  • MSA: 09:00:00 + 120q
  • MLD: 09:00:20 + 120q
  • LMU: 09:00:40 + 120q

Each ledger row’s local time is corrected to reference UTC before comparing.

Authentic frame selection

Only LN-17 / ARC-26251F rows are in scope. For each frame q, a payload is authentic when the same payload appears at all three stations at their expected corrected reference times, and the row’s check matches the recomputed CHECK integrity.

This cleanly separates the three classes of bad rows:

  • Corrupted rows — wrong payload / wrong check (e.g. 32Q2P6, 6ZA4TF, ME3OF5, ME3JF5).
  • Delayed / mistimed rows — valid checksum but logged 65 s off the expected slot (e.g. several duplicates at the wrong second).
  • Forged rows — valid CHECK but 185–199 s late (e.g. LOG-016, LOG-004, LOG-031), which the notes warn cannot be trusted on CHECK alone.

The surviving authentic payloads, one per frame:

Decoding

Wire symbols: ABCDEFGHIJKLMNOPQRSTUVWXYZ234567
Plain symbols: ABCDEFGHIJKLMNOPQRSTUVWXYZ01234

For frame q, previous = (seed + q) mod 32, and at position i (1–6):

transmitted_index = (plain_index + seed + 7q + 3i + previous) mod 32
previous = transmitted_index

Decoding reverses this. CHECK = (sum(i * transmitted_index_i) + 17q + 9*seed) mod 251.

Frame Wire Plain
00 KRN6TY COASTA
01 32Q2O6 L TELE
02 6WA4TF GRAPH
03 VHJFBD VERIFI
04 ME3DF5 ED NEX
05 D6PEJE T STAT
06 AREMJY ION RA
07 DVQ7RG DIO

Concatenating in frame order and stripping only terminal padding spaces:

COASTAL TELEGRAPH VERIFIED NEXT STATION RADIO

Flag computation

Per the FINAL PROOF in operator_notes.txt:

R1 = first 12 uppercase hex of SHA-256(
     LS01|TEAM_ID|PRIVATE_NONCE|CASE_ID|MESSAGE|STATION_CODES)

With TEAM_ID=SHARED, PRIVATE_NONCE=8332EE3D89FAD515, CASE_ID=ARC-26251F, the message above, and stations MSA-MLD-LMU:

LS01|SHARED|8332EE3D89FAD515|ARC-26251F|COASTAL TELEGRAPH VERIFIED NEXT STATION RADIO|MSA-MLD-LMU

SHA-256: 8ec5c9ebf7c3a3050c7005527c65e53e096527029e7956ab3066158d412e9e7b
R1: 8EC5C9EBF7C3

Flag

FLAG{REDACTED}

R1 (8EC5C9EBF7C3) is retained for Chapter 2.

Vulnerability Analysis

This is a forensics / protocol-reconstruction puzzle illustrating that checksum validity does not imply authenticity, and that clock skew can make genuine records appear inconsistent.

Mitigations

  • Corroborate message authenticity across independent stations (TX at source + RX at destination) rather than trusting a single copy or checksum alone.

  • Correct for clock skew using authoritative time sources before comparing timestamps across systems.

  • Authenticate each record with a keyed integrity check and verify the full chain of custody, not just individual record checksums.

  • Never trust a checksum alone: a valid CHECK can still belong to a forged or delayed relay copy. Corroborate via route, case ID, independent stations, and corrected timing.

  • Calibrate every clock to a common reference before comparing timestamps; local logs are not synchronized.

  • Enumerate candidate lines from primary sources before decoding; the wrong line ID yields the wrong seed and garbage plaintext.

  • Verify a custom encoder against any provided known-example vector before trusting its output.

Techniques

  • Custom Encoding
  • Clock Synchronization
  • Historical Research