Skip to content

Repository files navigation

canary-scan

pypi python build tests license

Scan document data-sources for canaries, trackers, web beacons, and per-recipient fingerprints before interacting with supplied datasets.

When you receive a large document dump from an external party — a leak, legal disclosure, or investigation — those files and documents can contain deliberate or indirect canaries: tracking pixels, embedded JavaScript, remote template links, steganographic watermarks, or per-recipient metadata fingerprints that phone home the moment a file is opened.

canary-scan inspects files without opening it in its native viewer, extracting and analysing raw structure, metadata, embedded objects, and near-duplicate fingerprints to surface anything that may reveal to an external party that the data-source is being examined.

Full documentation: psaintelligence.github.io/canary-scan


Quick Start (Docker)

The recommended way to run canary-scan is via Docker, as the image bundles all required system utilities and dependencies:

# Run the scan using the GitHub Container Registry image
docker run --rm \
  -v /mnt/datasource:/data:ro \
  -v $(pwd)/canary-scan-out:/output \
  ghcr.io/psaintelligence/canary-scan:latest scan /data -o /output

# Review findings
jq '.[] | select(.severity=="critical")' canary-scan-out/canary-scan-report.json

Quick Start (pipx)

If you prefer to run canary-scan directly on your host machine:

# 1. Install canary-scan
pipx install canary-scan

# 2. Install required system dependencies (Ubuntu 24.04 example)
sudo apt install libimage-exiftool-perl qpdf poppler-utils mupdf-tools \
    ripgrep unzip p7zip-full

# 3. Run the scan
canary-scan scan /mnt/datasource

Detection pipeline

Seven sequential stages: inventory → metadata → remote-refs → embedded → stego → uniqueness → report

Each stage writes a JSONL artefact to .canary-scan/. Run canary-scan --guide for a concise cheat sheet.


Supply-chain safety

canary-scan pins third-party Python dependencies to releases no newer than 14 days via uv's exclude-newer setting in pyproject.toml. This bounds the window during which a compromised upstream package could enter the resolved lockfile. The Didier Stevens bundled scripts are pinned at build time via make vendor; their provenance (commit SHA + version + date) is recorded in src/canary_scan/bundled/VERSIONS.txt.


License

Apache-2.0. Bundled third-party scripts (pdfid, pdf-parser, rtfdump) are BSD 2-Clause — see src/canary_scan/bundled/README.md.

About

Scan document data-sources for canaries, trackers, web beacons, and per-recipient fingerprints before interacting with supplied datasets.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages