@viraatdas/rudder@2.14.57
A Claude Code-style terminal app for running coding agents with worktree-isolated runs.
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThis is a dangerous, source-confirmed credential-upload capability reached by explicit cloud commands, not a proven silent install-time theft chain. Its automatic postinstall installation of a separate tool further increases supply-chain risk.
Decision evidence
public snapshot- Its postinstall hook automatically installs jj through Homebrew or Cargo when jj is absent.
- The explicit cloud-launch command archives a repository snapshot and posts its base64 contents to the vendor cloud service.
- Cloud snapshots include agent credentials, cloud-provider configuration, shell files, and package-manager configuration by default.
- The snapshot code captures a broad allowlist of secret environment variables, including OpenAI, AWS, GitHub, npm, database, and Slack credentials.
- On macOS, cloud snapshot creation reads Claude credentials from the Keychain and stages them for upload.
- Credential collection and upload are reached through explicit cloud commands rather than during package installation or import.
- The manifest has no runtime dependency on another release of this package.
- The postinstall command uses fixed command arguments and does not download or execute package-controlled remote code.
Source & flagged code
21 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgSource file is highly similar to a previously finalized malicious package; route for source-aware review.
dist/util.jsView on unpkgFindings
Affected versions and remediation
This report applies to @viraatdas/rudder@2.14.57.
See all versions of @viraatdas/rudder for other recorded verdicts.
- Review the evidence and your use of @viraatdas/rudder@2.14.57 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.