This guide publishes benchmark evidence so readers can verify paper claims without any AWS credentials.
s3://...links usually require IAM access- Public release assets are easier for reviewers and readers
- Checksums make integrity verifiable
./scripts/prepare_public_artifact.sh \
--input-dir path/to/generated/kms_validation_<timestamp> \
--name kms_validation_<timestamp>.tar.gzOutputs in artifacts/public/:
kms_validation_<timestamp>.tar.gzkms_validation_<timestamp>.tar.gz.sha256kms_validation_<timestamp>.tar.gz.manifest.txt
The script scans for common sensitive markers (ARNs, instance IDs, access keys, IPv4, account-ID contexts) and fails by default if found.
Authenticate once:
gh auth login -h github.comUpload to an existing tag (example v1.0.0):
./scripts/publish_public_artifact.sh \
--tag v1.0.0 \
--artifact artifacts/public/kms_validation_<timestamp>.tar.gz- Release page URL (stable):
https://github.com/cyntrisec/EphemeralML/releases/tag/v1.0.0 - Artifact file name:
kms_validation_<timestamp>.tar.gz - SHA-256 checksum:
<sha256-from-generated-artifact>
After download:
sha256sum -c kms_validation_<timestamp>.tar.gz.sha256Expected output:
...: OK
If you need archival citation quality:
- Connect the GitHub repo to Zenodo.
- Archive the release containing the artifact.
- Add DOI to paper provenance and README.