This looks like a three part challenge - git forensics, a classified PDF, and sentence embeddings. You need to find two variables X and Y to build the flag.
The challenge references the 2013 Snowden disclosures which points you to the snowden archive on github (iamcryptoki/snowden-archive). The target is the BULLRUN classification guide PDF from September 5 2013. The gotcha here is that the challenge wants the commit SHA for that specific file, not the latest commit on the repo… these are different. You scope git log to just that file path:
git clone https://github.com/iamcryptoki/snowden-archive
cd snowden-archive
git log -- documents/2013/20130905-theguardian__bullrun.pdf
# commit 7d88323521194ed8598624dc3a932930debdde1dFirst 7 chars of that SHA give you X = 7d88323.
Next you open the PDF itself. The challenge says Appendix A but the ECI list is actually in the Remarks column of row A.2 on page 2:
APERIODIC, AMBULANT, AUNTIE, PAINTEDEAGLE, PAWLEYS, PITCHFORD, PENDLETON, PICARESQUE, PIEDMONTSecond entry, 8 letters, which refers to AMBULANT, normalized to ambulant.
Last part, run that word through all-MiniLM-L6-v2 and take index 0 of the output vector rounded to 4 decimal places. The model is deterministic so same input always gives the same number. Y = 0.0245.
The flag -
apoorvctf{7d88323_0.0245}code here
