@viraatdas/rudder@2.14.55
A Claude Code-style terminal app for running coding agents with worktree-isolated runs.
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThe package contains an automatic install hook and substantial credential-upload functionality. Static source shows the latter is user-invoked and authenticated, so it is a dangerous capability rather than confirmed covert malware.
Decision evidence
public snapshot- The install hook automatically installs the jj command through Homebrew or Cargo when it is absent.
- Cloud snapshots include default paths holding Claude, Codex, GitHub, AWS, and other cloud credentials.
- A cloud launch sends the base64-encoded snapshot to the Rudder control plane.
- The explicit secrets-sync command reads allowlisted local files and sends their encoded contents to the cloud vault.
- Credential transfer is reached through explicit cloud commands and requires a cloud login.
- The remote bootstrap command is printed and only runs over SSH when a separate opt-in environment variable is set.
- No self-dependency or hidden install-time credential collection was found.
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@2.14.55.
See all versions of @viraatdas/rudder for other recorded verdicts.
- Review the evidence and your use of @viraatdas/rudder@2.14.55 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.