@viraatdas/rudder@3.1.18
A TUI for managing coding agents: run Claude Code, Codex and opencode side by side, each in its own isolated jj workspace.
AI Security Review
scanned 6d ago · by lpm-firewall-aiSnapshot uploads conditionally forward saved Claude credentials to Rudder Cloud without a credential-specific confirmation. The explicit cloud operation and vault gate make this a warning-level credential exposure risk, not sufficient evidence for a publish block.
Decision evidence
public snapshot- When creating a cloud snapshot, the code can read Claude’s saved macOS Keychain token, stage it in the snapshot, include credentials unless the cloud vault is active, and POST the archive to Rudder Cloud.
- The snapshot path is reached by the explicit cloud snapshot command or when a repository without a GitHub origin falls back to uploading a snapshot; Rudder Cloud is the default recipient.
- The other inspected credential flows query Claude at its service endpoint and Codex at chatgpt.com; cloud account-parity checks compare local identities with Rudder home state.
- Other flagged integration locations read Claude conversation files and a local Codex model cache; the board uses a loopback endpoint, and the inspected parsing and redaction helpers do not show credential export.
Source & flagged code
21 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time source downloads a native archive from a fixed external host without transport verification, extracts it, and installs an executable payload.
package.jsonView on unpkg · L1Findings
Affected versions and remediation
This report applies to @viraatdas/rudder@3.1.18.
See all versions of @viraatdas/rudder for other recorded verdicts.
- Review the evidence and your use of @viraatdas/rudder@3.1.18 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.