registry / @viraatdas/rudder / 3.2.1
@viraatdas/rudder@3.2.1
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 5d ago · by lpm-firewall-aiSource inspection found the postinstall hook performs Jujutsu setup and the inspected credential flows target their corresponding services or compare account identities. The scanner’s similarity hints did not establish malicious behavior in the reviewed files.
Static reason
One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review
Trigger
Package installation invokes the postinstall hook; other reviewed routines run when their CLI, cloud, usage, board, or improvement features are used.
Impact
No confirmed attack surface was established in the inspected flows.
Mechanism
The hook invokes the Jujutsu helper. Inspected usage code sends a local Codex token to the matching ChatGPT usage endpoint, and cloud parity compares local and cloud account identities.
AI rationale
Source inspection found the postinstall hook performs Jujutsu setup and the inspected credential flows target their corresponding services or compare account identities. The scanner’s similarity hints did not establish malicious behavior in the reviewed files.
Network endpoints1
chatgpt.com/backend-api/wham/usage
Decision evidence
public snapshotAI called this Clean at 93.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- package.json runs a postinstall hook; scripts/postinstall.mjs skips root installs and attempts Jujutsu setup through the package’s own checksum-pinned helper.
- The cloud account comparison reads local identities and the cloud home’s credential status; its cloud command reports differences and suggests fixes rather than exporting local tokens.
- Claude credential snapshotting writes a credential file into a staged cloud home, and Claude conversation and Codex model discovery read their respective local agent data.
- Codex usage reads the local Codex access token and sends it to chatgpt.com’s usage endpoint; the inspected caller gathers usage reports for the local board.
- The inspected improvement collector describes redacted session records, task summaries redact secret-like tokens, the feedback description excludes prompts and file contents, and the board uses loopback request checks.
- Required source paths were inspected; their cited excerpts show local runtime helpers, redaction, scheduling, backend/run management, version checking, and Jujutsu setup. No concrete attack behavior was identified.
Evidence against
- The package has a postinstall hook and contains credential-handling and agent-data code, but the inspected flows are consistent with Jujutsu installation, cloud setup, local usage reporting, and agent context discovery.
Source & flagged code
21 flagged · loading sourcepackage.jsonView file
•scripts.postinstall = node scripts/postinstall.mjs
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg1scripts.postinstall = node scripts/postinstall.mjs
Install-time code downloads an unverified remote native archive, stages it locally, activates an executable path, and exposes it to process execution.
L1: import { spawnSync } from "node:child_process";
L2: import { createHash, randomUUID } from "node:crypto";
...
L10: // ~/.rudder/bin/jj/<version>/ and puts that directory on PATH for itself and
L11: // everything it starts: the native dashboard, workers and agents all inherit
L12: // it. A jj the user installed (brew, cargo, a distro package) always wins.
...
L15: // (`git worktree add --orphan`), and stock macOS and Debian 12 ship git 2.39,
L16: // so every isolated agent would fail to start on exactly the machines that have
L17: // nothing installed. 0.40.0 is also what the cloud worker
High
Unverified Remote Native Payload Install
Install-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 · L1•scripts.postinstall = node scripts/postinstall.mjs
Findings
17 High6 Medium5 Low
HighInstall Time Lifecycle Scriptspackage.json
HighCredential Redirect Persistencedist/cloud.js
HighUnverified Remote Native Payload Installpackage.json
HighTrigger Reachable Persistencedist/improve/schedule.js
HighSemantic Analysis Limiteddist/improve/collect.js
HighKnown Malware Source Similaritydist/usage.js
HighKnown Malware Source Similaritydist/util.js
HighKnown Malware Source Similaritydist/backends.js
HighKnown Malware Source Similaritydist/board/daemon.js
HighKnown Malware Source Similaritydist/handoff.js
HighKnown Malware Source Similaritydist/run-manager.js
Affected versions and remediation
This report applies to @viraatdas/rudder@3.2.1.
See all versions of @viraatdas/rudder for other recorded verdicts.
- Review the evidence and your use of @viraatdas/rudder@3.2.1 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.