Skip to content

Data Exfiltration: flag snapshot, export and batch secret-read actions #654

Open

Description

@kmcquade

The Data Exfiltration finding flags five actions (READ_ONLY_DATA_EXFILTRATION_ACTIONS in
cloudsplaining/shared/constants.py): s3:GetObject, ssm:GetParameter, ssm:GetParameters,
ssm:GetParametersByPath and secretsmanager:GetSecretValue.

The DataExfiltration risk definition in the same file also says rds:CopyDBSnapshot and
rds:CreateDBSnapshot "can be used to exfiltrate RDS database contents", but neither is in the list,
so a policy that allows them on * is not flagged.

Proposed additions

Secret and parameter reads (the same shape as the existing entries):

Action Why
secretsmanager:BatchGetSecretValue Returns the values of up to 20 secrets in one call.
ssm:GetParameterHistory Returns a parameter's past values, decrypted when WithDecryption is set.

Snapshot, copy and export. Each puts a full copy of a data store where it can be shared with
another account or downloaded:

Action Why
rds:CreateDBSnapshot, rds:CopyDBSnapshot Already named in the risk definition.
rds:CreateDBClusterSnapshot, rds:CopyDBClusterSnapshot The Aurora cluster equivalents.
rds:StartExportTask Exports a DB or cluster snapshot to S3.
ec2:CreateSnapshot, ec2:CreateSnapshots, ec2:CopySnapshot EBS volume snapshots and copies.
ec2:CreateInstanceExportTask Exports an instance's disks to S3.
ec2:ExportImage Exports an AMI to S3.
dynamodb:ExportTableToPointInTime Exports a whole table to S3.

Design question

These are Write-level actions, so they don't fit a list named READ_ONLY_…. Two options:

  1. A second list, reported under the same Data Exfiltration finding.
  2. A separate finding type for "copies data out of its store".

Related: #272 (more read actions, closed) and #624 (include-actions doesn't reach the Data
Exfiltration check).

Activity

  1. devteamaegis commented on Nov 4, 2026

    @devteamaegis

    I opened #668 for this: the snapshot/export actions and ssm:GetParameterHistory are reported under the existing Data Exfiltration finding. I left secretsmanager:BatchGetSecretValue out (it can only be granted on * and AWS also requires GetSecretValue per secret, so it would flag correctly scoped policies); the PR explains this, and I'm happy to add it if you prefer.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions