agentleFS
acprkacprk/crypto-research-agents/skills/anonymize-check/SKILL.md

anonymize-check

Skill15 starsChanged 9 days agoMITView on GitHub

Use before every double-blind submission, supplementary upload, anonymous artifact mirror or rebuttal attachment. Scans .tex/.bib/code/zip/PDF (text, metadata, embedded paths, compressed object streams) for author names, first-person self-citations, personal code-host URLs, e-mails, local paths, acknowledgements and funding lines, using a configurable deny-list.

The file

Files
SKILL.md

Anonymize check

Script

scripts/anon_check.py (standard library only; uses PyMuPDF or poppler if present).

# 1. keep a private deny-list (never commit a filled-in one publicly)
cp skills/anonymize-check/scripts/denylist.example.txt paper/.denylist.txt   # then edit

# 2. scan sources, the built PDF, the supplementary zip, the artifact directory
python3 skills/anonymize-check/scripts/anon_check.py paper/ paper/main.pdf \
        supplementary.zip artifact/ --deny-file paper/.denylist.txt \
        --allow-url 'github\.com/<third-party-org>/'

# 3. self-test (synthetic sample; must print PASS)
python3 skills/anonymize-check/scripts/anon_check.py --selftest

Exit status 1 if any HIGH finding. --json for machine-readable output, --min-severity MEDIUM to reduce noise.

What it flags

severityfinding
HIGHdeny-list term in text/code/filename/PDF bytes; e-mail; local path (/home/<u>, /Users/<u>, C:\Users\<u>); non-anonymous \author/\institute/\thanks/\email; acknowledgements section; PDF Author metadata; embedded figure path (PTEX.FileName) containing a username or deny term; .git inside an archive; URL containing a deny term
MEDIUMfirst-person self-citation ("our previous work", "we previously showed", "our paper~\cite"); personal code-host URL (github/gitlab/.github.io/...) not in the allow-list; funding statement; PDF Title/Subject/Keywords metadata; .git directory in an artifact folder; hits inside comments (downgraded one level)
LOWdeny term inside .bib (allowed: self-citations stay in the bibliography with real authors, referred to in the third person); "camera-ready"/"de-anonymised" wording; Creator/Producer metadata (informational)

Manual checks the script cannot do

  • Self-citations are written in the third person ("Example et al. [12] showed") and critiqued like any other work; not omitted, not "anonymised" in the bib.
  • No "blinded" citation that is obviously the authors' own when a third-person citation would do (follow the venue's policy; IACR generally prefers third person).
  • Supplementary/artifact README avoids "our prior paper", "same codebase as", "inherited from our earlier implementation".
  • Anonymous mirrors (e.g. anonymous.4open.science) are refreshed after the last push, and each file URL returns HTTP 200 (a private source repository can make the mirror return 403). Check with curl -sI.
  • Figures: no screenshots with usernames, terminal prompts, or window titles.
  • Experiment logs in the artifact: hostnames, IPs, usernames stripped.
  • Overleaf/Git history is not shared with reviewers.
  • Submission form fields (conflicts, abstract) are separate from the PDF; the PDF itself must be anonymous.

Remediation snippets

% Empty PDF metadata (hyperref)
\hypersetup{pdfauthor={},pdftitle={},pdfsubject={},pdfkeywords={},pdfcreator={},pdfproducer={}}
% Avoid embedding figure source paths (pdfTeX)
\pdfsuppressptexinfo=-1   % place in the preamble (pdfTeX >= 1.40.15)
# strip metadata from a finished PDF (qpdf/exiftool if available)
exiftool -all:all= -overwrite_original main.pdf && qpdf --linearize main.pdf main_clean.pdf
# zip an artifact without VCS data
git archive --format=zip -o artifact.zip HEAD   # or: zip -r artifact.zip artifact -x '*.git*'

Record the final clean run (command + summary line + PDF sha256) in SUBMISSION.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.