Close Open Privacy Scan

bolt Snapshot: commit b6d5084
science engine v1.23
schedule 2026-07-25T20:08:12.221653+00:00

verified_user Possible application data leak

Potential data exfiltration identified in application code.

App Privacy Score

22 /100
High privacy risk — possible application leak

High risk · 1107 finding(s)

Based on: 1 first-party package(s) · 0/0 deps analyzed

Dependency score: 100 (Low risk)

bar_chart Score Breakdown

pii_flow −60
egress −15
env_fs −3

list Scan Summary

0 high 6 medium 1101 low
First-party packages: 1
Dependency packages: 0
Ecosystem: python

swap_horiz Potential data exfiltration in application code

External domains: app.neptune.aiarxiv.orgaxtkn4xl5cip.objectstorage.us-phoenix-1.oci.customer-oci.combugs.python.orgcdn-datasets.huggingface.cocdn-media.huggingface.cocdn.britannica.comcircleci.comclear.mlclip-cn-beijing.oss-cn-beijing.aliyuncs.comconversationhub.blob.core.windows.netdagshub.comdiscuss.huggingface.codiscuss.pytorch.orgdl.fbaipublicfiles.comdocs.neptune.aidocs.nvidia.comdocs.python.orgdocs.pytorch.orgdocs.ray.iodocs.swanlab.cndocs.wandb.aidownload.openmmlab.comdvc.orgen.wikipedia.orgessentia.upf.edufarm4.staticflickr.comflyte.orggiou.stanford.edugist.github.comgithub.comguillaumejaume.github.iohf.cohips.hearstapps.comhub-ci.huggingface.cohuggingface.coi.postimg.ccimages.cocodataset.orgimages.pexels.comlayoutlm.blob.core.windows.netmistralai.github.ionineplanets.orgnlp.stanford.eduobject.pouta.csc.fioptuna.readthedocs.iopaddle-model-ecology.bj.bcebos.compbs.twimg.compraeclarumjj3.github.ioproceedings.neurips.ccprojects4jw.blob.core.windows.netpublicgit.blob.core.windows.netpypi.orgpytorch.orgqwenlm.github.ioraw.githubusercontent.comrrc.cvc.uab.essafetensors-convert.hf.spacestorage.googleapis.comsuno-ai.notion.siteswanlab.cnthumbs.dreamstime.comtwitter.comuploads4.wikiart.orgwww.comet.comwww.cs.cmu.eduwww.deepmind.comwww.dropbox.comwww.dvc.orgwww.ilankelman.orgwww.kxan.comwww.mlflow.orgwww.nature.comwww.tensorflow.orgwww.wandb.com

medium first-party (python) PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
  1. 1sourcerepo/src/transformers/integrations/integration_utils.py:2515
  2. 2sinkrepo/src/transformers/integrations/integration_utils.py:2517
medium first-party (python) PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
  1. 1sourcerepo/src/transformers/models/regnet/convert_regnet_seer_10b_to_pytorch.py:236
  2. 2sinkrepo/src/transformers/models/regnet/convert_regnet_seer_10b_to_pytorch.py:236
medium first-party (python) A credential read from the environment/filesystem is applied as authorization on the same outbound request (auth header). This is intentional authentication, not unexpected data exfiltration.
  1. 1sourcerepo/utils/notification_service.py:1079
  2. 2sinkrepo/utils/notification_service.py:1084
medium first-party (python) A credential read from the environment/filesystem is applied as authorization on the same outbound request (auth header). This is intentional authentication, not unexpected data exfiltration.
  1. 1sourcerepo/utils/notification_service.py:1079
  2. 2sinkrepo/utils/notification_service.py:1098
medium first-party (python) PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
  1. 1sourcerepo/utils/process_bad_commit_report.py:37
  2. 2sinkrepo/utils/process_bad_commit_report.py:151
medium first-party (python) PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
  1. 1sourcerepo/utils/update_pr_ci_dashboard_recap.py:331
  2. 2sinkrepo/utils/update_pr_ci_dashboard_recap.py:350

</> First-Party Code

first-party (python)

python first-party
medium pii_flow production #1d5c7255072af9fc PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
repo/src/transformers/integrations/integration_utils.py:2517 · flow /tmp/closeopen-vlu0356z/repo/src/transformers/integrations/integration_utils.py:2515 → /tmp/closeopen-vlu0356z/repo/src/transformers/integrations/integration_utils.py:2517
            logger.debug(f"[Kubeflow] Token file not found: {token_path}")

PII-bearing data is written to a log/print sink — it stays in-process and does not leave the application, but logged PII is still a privacy concern.

Fix: Avoid logging user identifiers; redact or omit PII from log/print statements.

medium pii_flow production #dd167469f036c654 PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
repo/src/transformers/models/regnet/convert_regnet_seer_10b_to_pytorch.py:236 · flow /tmp/closeopen-vlu0356z/repo/src/transformers/models/regnet/convert_regnet_seer_10b_to_pytorch.py:236 → /tmp/closeopen-vlu0356z/repo/src/transformers/models/regnet/convert_regnet_seer_10b_to_pytorch.py:236
        logger.info(f"Token is {os.environ['HF_TOKEN']}")

PII-bearing data is written to a log/print sink — it stays in-process and does not leave the application, but logged PII is still a privacy concern.

Fix: Avoid logging user identifiers; redact or omit PII from log/print statements.

medium pii_flow production #b40a25aa36946eef A credential read from the environment/filesystem is applied as authorization on the same outbound request (auth header). This is intentional authentication, not unexpected data exfiltration.
repo/utils/notification_service.py:1084 · flow /tmp/closeopen-vlu0356z/repo/utils/notification_service.py:1079 → /tmp/closeopen-vlu0356z/repo/utils/notification_service.py:1084
        ci_details = requests.get(ci_detail_url, headers=github_headers).json()

A credential (read from the environment/filesystem, or parsed from the request URL) is applied as authorization on the same outbound request. This is intentional authentication to the service the credential belongs to, not unexpected data exfiltration.

Fix: Confirm the destination is the credential's own service; scope the credential and avoid logging it. No action if this is the intended authenticated API call.

medium pii_flow production #2b4cf3d2eff453e6 A credential read from the environment/filesystem is applied as authorization on the same outbound request (auth header). This is intentional authentication, not unexpected data exfiltration.
repo/utils/notification_service.py:1098 · flow /tmp/closeopen-vlu0356z/repo/utils/notification_service.py:1079 → /tmp/closeopen-vlu0356z/repo/utils/notification_service.py:1098
            ci_details = requests.get(ci_detail_url, headers=github_headers).json()

A credential (read from the environment/filesystem, or parsed from the request URL) is applied as authorization on the same outbound request. This is intentional authentication to the service the credential belongs to, not unexpected data exfiltration.

Fix: Confirm the destination is the credential's own service; scope the credential and avoid logging it. No action if this is the intended authenticated API call.

medium pii_flow production #9c9f78e39333d5be PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
repo/utils/process_bad_commit_report.py:151 · flow /tmp/closeopen-vlu0356z/repo/utils/process_bad_commit_report.py:37 → /tmp/closeopen-vlu0356z/repo/utils/process_bad_commit_report.py:151
    print(report)

PII-bearing data is written to a log/print sink — it stays in-process and does not leave the application, but logged PII is still a privacy concern.

Fix: Avoid logging user identifiers; redact or omit PII from log/print statements.

medium pii_flow production #7f3a1df8cc630b91 PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process.
repo/utils/update_pr_ci_dashboard_recap.py:350 · flow /tmp/closeopen-vlu0356z/repo/utils/update_pr_ci_dashboard_recap.py:331 → /tmp/closeopen-vlu0356z/repo/utils/update_pr_ci_dashboard_recap.py:350
    print(f"Matched PR #{pr['number']}: {pr['html_url']}")

PII-bearing data is written to a log/print sink — it stays in-process and does not leave the application, but logged PII is still a privacy concern.

Fix: Avoid logging user identifiers; redact or omit PII from log/print statements.

expand_more 1101 low-confidence finding(s)
low env_fs Environment-variable access. 293 locations
low env_fs Filesystem access. 765 locations
low pii_flow PII-bearing data is written to a log/print sink. Logged PII is a privacy concern even when it does not leave the process. Non-production path — not application runtime. 6 locations
low egress Outbound request to a variable or assembled URL on a network client. Review what data is sent to this destination. 33 locations
low egress Hardcoded external endpoint. Review what data is sent to this destination. 4 locations