November 19, 2026
From a Blank PDF to a Manhattan Address: Solving the “Counterintelligence” Challenge on the OSINT…
Document forensics, social media pivoting, audio steganography, Morse code and SSTV decoding

By Chiranjeevi C
4 min read
Spoiler warning:_ this write-up contains the full solution to a challenge on ctf.osint.industries. Try it yourself first._
TL;DR
The goal of this challenge was to find a location. The only starting material was a PDF that looked empty. By extracting its hidden text layer I found a Mastodon account, followed a trail through two accounts and two audio files, decoded an emoji cipher and Morse code, and finally decoded a Slow-Scan Television (SSTV) signal that produced a QR code containing the answer.
- Platform: ctf.osint.industries
- Challenge: Counterintelligence
- Objective: find the meeting location
- Skills: document forensics, social media OSINT, audio analysis, cipher decoding, SSTV
- Tools: exiftool, pdftotext, Mastodon, Audacity, Sonic Visualizer, QSSTV
The Challenge
Counterintelligence had found a suspicious PDF named "nothing to see here" on a public PDF-sharing website. Investigators suspected it hid covert communication between two foreign agents arranging a meeting. The task was to find where they were supposed to meet.
The rules were strict: passive OSINT only, no interaction with people, places, or accounts, and no authentication or recovery attempts. The flag format was OSINT{Location}.
When I opened the PDF link, there was nothing visible, which in an OSINT challenge usually means the information is hidden, not absent.
Step 1: Interrogating the "empty" PDF
I downloaded the file instead of relying on the browser view and checked it in two layers. First the metadata: exiftool returned nothing useful, but one detail stood out: the document language was set to French (fr). Second, the content: treating that tag as a hint that there was real text inside, I converted the PDF to text with pdftotext.
exiftool challenge.pdf
pdftotext challenge.pdf -
The output was short: the phrase "NOTHING TO SEE HERE !" followed by Hegev45373 @ m4st0d0n. The handle was written in leetspeak ("m4st0d0n" is Mastodon), which told me exactly which platform and account to check.
Takeaway: a document that looks blank is not necessarily empty. Always compare what a file shows with what it contains.
Step 2: Pivoting to Mastodon
The extracted text led to the account @Hegev45373 on mastodon.social. One post carried an audio attachment named c09388da46a0dd58.mp3 and a caption made only of emoji.
Step 3: The first audio file, a dead end that wasn't
Audio files in CTFs often hide text or images in the frequency domain, so I looked at the MP3 as a spectrogram. It displayed a message: "no, not here…"
That looked like a dead end, but it was actually a clue: the answer was not in this file. The same post had another element I hadn't used yet, the emoji caption.
Step 4: Decoding the emoji cipher
The caption was a sequence of emoji that didn't form a sentence. After considering that it might be an emoji-based cipher, I decoded it. The output was the URL of a second Mastodon account: @rawepe5298.
Takeaway: when a result says "not here," go back and re-read every element of the source, including captions, alt text and timestamps.
Step 5: The second account and a stubborn audio file
The new account had its own post with an audio file, dd2d5990d57cf4f8.mp3. It is flagged as automated, has no followers, and has exactly one post, so nothing else distracts from it. Opened in Audacity, the spectrogram was not readable.
As before, I turned to the caption. This time it was Morse code:
.- . — . . — . .-. — — .- -.-. …. .. -. — .
Decoded, it reads APPROACHING. That told me two things: the audio file really did contain the location, and I simply wasn't viewing it correctly.
Step 6: Changing tools
Rather than keep adjusting Audacity, I switched to Sonic Visualizer, whose spectrogram layer gave a much clearer picture. The pattern looked like a structured, telephone-style signal, which pointed to a digital communication protocol rather than speech or hidden text.
After testing a few possibilities, I identified it as SSTV (Slow-Scan Television), a method of transmitting still images as audio tones, historically used by amateur radio operators.
Takeaway: if one tool gives you noise, try another before assuming the data isn't there. Visualization settings matter.
Step 7: Decoding the SSTV signal
I played the audio into QSSTV, a software SSTV decoder. The output image was a QR code. Scanning it with my phone gave the final message:
Please Don't Tell, 133 St Marks Place, NY 10009
Answer
The location is OSINT{Please_Don't_Tell,133_St_Marks_Place,_NY_10009}.
The whole solution at a glance
Lessons learned
- Look at content, not just appearance. Hidden text layers are easy to miss in a browser.
- Metadata is a hint even when it is not the answer. The French language tag nudged me toward text extraction.
- Every element of a post is evidence. The captions mattered as much as the audio files.
- Tool choice matters. Audacity failed, whereas Sonic Visualizer succeeded.
- Recognize the protocol. Identifying SSTV was the turning point.
Skills demonstrated
- Document and metadata analysis
- Social media investigation and account pivoting
- Audio steganography and spectrogram analysis
- Cipher decoding (emoji cipher, Morse code)
- Signal identification and SSTV decoding
- Methodical problem solving through dead ends
A note on ethics
This investigation was conducted within a controlled Capture The Flag (CTF) environment using fictional accounts and synthetic artifacts; real-world OSINT should always remain strictly passive and respect platform terms of service.
If you enjoy challenges like this, try ctf.osint.industries yourself. Feedback and alternative approaches are welcome in the comments.