A compromised medical-imaging device leaked patien...

作成日: 2026年11月8日

使用モデル GPT-6 Thinking by Chat01

質問

dicom_device.zip

A compromised medical-imaging device leaked patient data across a hospital network. The DICOM capture, device configuration, access ledger, SNMP alerts, and recovered exfiltration stream must be treated as corroborating evidence rather than independent answers.

Reconstruct the intrusion and recover its marker from the supplied offline evidence.

Flag format: CSCTF{8_lowercase_hex}

Flag derivation (serialization spec): the flag is sha256("<dicom_serial>:<snmp_iso_timestamp>:<leak_count>:<anomaly_count>:<manifest_seed>")[:8], lowercase hex.

Serialization details:

dicom_serial = the device serial string from dicom_device.xml.
snmp_iso_timestamp = the exact ISO 8601 timestamp from the anomalous event in snmp_trap.log, with T and trailing Z.
leak_count = number of leaked patient records as decimal integer.
anomaly_count = number of anomalous C-STORE operations in the pcap as decimal integer.
manifest_seed = the calibration seed that must agree between dicom_device.xml and snmp_trap.log, as decimal.
Fields joined as strings with : between them. No binary packing. No other fields.

回答

I’ll inspect the archive, correlate the device configuration, access records, SNMP alert, DICOM traffic, and recovered exfiltration stream, then compute the flag from the five specified fields. I’ll check the counts against the evidence rather than relying on any single file.

思考

Inspected cyber forensics archives, DICOM records, logs, and SNMP PCAP tools

回答

The device XML identifies serial MEDSCAN-XR400-2019, and its manifest seed 8842 matches the SNMP boot report. The patient ledger contains 90 distinct records marked Leaked. I’m checking the packet capture and alert timeline to isolate the correct event timestamp and C-STORE anomaly count.

思考

Examining capture anomalies

I’m correlating DICOM traffic, external transfers, SNMP alerts, and exfiltration artifacts. The likely pattern is 50 routine verification messages, 20 C-STORE anomalies, 90 leaked records, and one precisely timed alert to identify.

Examined capture anomalies

思考

Resolving the indicators

I’m treating the first September 1 alert as the correlated event because the capture aligns with that date. The evidence supports 20 anomalous C-STORE operations, 90 leaked records, repeated serial text, and hash 86a4ae95.

Resolved the indicators

思考

Analyzed network evidence and computed the recovered exfiltration flag

Worked for 1m 9s

回答

Text version

Flag recovered

CSCTF{86a4ae95}

[Copy flag]

1. Recovered flag components

FieldVerified valueEvidence
dicom_serialMEDSCAN-XR400-2019Device XML, corroborated by recovered stream
snmp_iso_timestamp2026-09-01T03:15:00ZFirst anomalous SNMP alert on the capture date
leak_count9090 unique records marked Leaked
anomaly_count20Outbound DICOM C-STORE operations
manifest_seed8842XML and SNMP boot report agree

2. Intrusion reconstruction

  1. Exposed medical-imaging device

    The imaging device at 10.20.30.45 had unchanged default credentials, an exposed SNMP community string, disabled firewall protection, and unencrypted DICOM traffic.

  2. Abnormal DICOM transfers

    hospital_network.pcap contains 50 routine C-ECHO operations to the internal archive 10.20.40.10, followed by 20 C-STORE operations from the imaging device to external address 198.51.100.42 on TCP port 104.

  3. Network reconnaissance

    The capture records 30 SNMP packets originating from 198.51.100.42. The SNMP log separately reports anomalous bulk polling using the device's configured community string. The first corresponding alert is dated September 1, 2026, at 03:15:00Z. Later alerts repeat on subsequent days.

  4. Patient-data exfiltration

    The capture also shows 15 HTTP upload requests from 10.20.40.10 to 198.51.100.42:8443. The patient CSV identifies 90 distinct leaked records across Wings C and E. The recovered binary stream contains repeated serial fragments consistent with the device XML.

The evidence supports unauthorized DICOM transfers, SNMP reconnaissance, and outbound upload activity involving the same external endpoint.

3. SHA-256 verification

Exact serialized input:

text
MEDSCAN-XR400-2019:2026-09-01T03:15:00Z:90:20:8842

Full SHA-256 digest:

text
86a4ae9597961a7446d8d27dde6a80d07f0adec01cb973d21e1c677124a9bdf6

Taking the first eight lowercase hexadecimal characters yields:

CSCTF{86a4ae95}

Timestamp selection: The SNMP log contains repeated daily anomaly alerts. September 1 is selected because it is the date corroborated by the packet capture.

このQ&Aを共有