registry / Credential exfiltration / @viraatdas/rudder / 3.9.7

@viraatdas/rudder@3.9.7

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 3h ago · by lpm-firewall-ai

The source shows a product-aligned cloud workflow that deliberately transfers saved provider credentials for remote use, but does not obtain separate consent for that transfer. Because the recipient can be configured and the source does not show misuse, this warrants a warning for credential exposure rather than a malware block.

Static reason

One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review

Trigger

A user starts a cloud launch or snapshot operation; workspace attachment may upload a snapshot when one is needed and the cloud vault is not active.

Impact

Provider credentials may become accessible to the control plane and remote worker, exposing connected accounts if that service or its configuration is not trusted.

Mechanism

The package stages credential files and selected token environment variables into a workspace archive, then posts that archive to the configured control plane.

AI rationale

Evidence

dist/cloud.js~/.claude/.credentials.json~/.codex/auth.json~/.aws/credentials

Network endpoints1

rudder-cloud-control.fly.dev

Decision evidence

public snapshot

AI called this Suspicious at 84.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

Cloud snapshots can include saved agent and provider credentials, then upload the archive during a cloud workspace operation.
The default control plane is Rudder’s cloud service, but an HTTPS control-plane URL can be configured; the source has no separate approval gate for exporting those credentials.
Codex usage credentials are also sent to the intended ChatGPT usage endpoint.
Credential parity code compares local and cloud account identities; the board quota route serves a local API request.
The install hook sets up the workspace tool, whose release archives have pinned checksums.
The utility creates a filesystem lock directory for a JSON critical section.
Task summary redaction splits text into tokens and passes each token to a redaction helper.
The improvement-state timeout marks the child process as timed out and sends it SIGTERM.

Evidence against

Credential transfer is part of an explicitly invoked cloud workflow and supports signing agents into remote workspaces.
When a cloud secrets vault is active, the workspace attach path excludes local credentials from its snapshot.
The inspected source does not establish that transferred credentials are later misused.

Source & flagged code

24 flagged

package.jsonView file

•scripts.postinstall = node scripts/postinstall.mjs

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

1scripts.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

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg

dist/onboarding.jsView file

5// both rotate refresh tokens, and a shared copy signs one side out. L6: import { spawnSync } from "node:child_process"; L7: import readline from "node:readline/promises";

High

Child Process

Package source references child process execution.

dist/onboarding.jsView on unpkg · L5

dist/util.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/util.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/util.jsView on unpkg

1import { spawn, spawnSync } from "node:child_process"; L2: import { createHash, randomUUID } from "node:crypto"; ... L7: import readline from "node:readline/promises"; L8: import { stdin as input, stdout as output } from "node:process"; L9: const DEFAULT_PATH = [ ... L24: export function rudderHome() { L25: return path.resolve(process.env.RUDDER_HOME?.trim() || path.join(os.homedir(), ".rudder")); L26: } ... L65: const raw = await fsp.readFile(filePath, "utf8"); L66: return JSON.parse(raw); L67: } ... L160: const TOOL_INSTALL_HINTS = {

Low

Weak Crypto

Package source references weak cryptographic algorithms.

dist/util.jsView on unpkg · L1

dist/improve/schedule.jsView file

7Trigger-reachable persistence chain: scripts.start -> dist/index.js -> dist/main.js -> dist/improve/index.js -> dist/improve/schedule.js L7: function plistPath() { L8: return path.join(os.homedir(), "Library", "LaunchAgents", `${LAUNCHD_LABEL}.plist`); L9: } ... L49: export async function scheduleCommand(action) { L50: if (process.platform !== "darwin") { L51: throw new Error("rudder improve schedule currently supports macOS (launchd) only; run `rudder improve run` from cron/systemd on other platforms."); ... L59: // Re-bootstrap idempotently: boot out any prior registration first. L60: await runCommand("launchctl", ["bootout", `gui/${uid}/${LAUNCHD_LABEL}`], { allowFailure: true }); L61: const bootstrap = await runCommand("launchctl", ["bootstrap", `gui/${uid}`, plistPath()], {

High

Trigger Reachable Persistence

A manifest entrypoint or package-local install chain reaches persistence behavior.

dist/improve/schedule.jsView on unpkg · L7

7function plistPath() { L8: return path.join(os.homedir(), "Library", "LaunchAgents", `${LAUNCHD_LABEL}.plist`); L9: } ... L49: export async function scheduleCommand(action) { L50: if (process.platform !== "darwin") { L51: throw new Error("rudder improve schedule currently supports macOS (launchd) only; run `rudder improve run` from cron/systemd on other platforms."); ... L59: // Re-bootstrap idempotently: boot out any prior registration first. L60: await runCommand("launchctl", ["bootout", `gui/${uid}/${LAUNCHD_LABEL}`], { allowFailure: true }); L61: const bootstrap = await runCommand("launchctl", ["bootstrap", `gui/${uid}`, plistPath()], {

Medium

Install Persistence

Source writes installer persistence such as shell profile or service configuration.

dist/improve/schedule.jsView on unpkg · L7

assets/skills/google-cloud/scripts/gcp.mjsView file

108async function refresh(cred) { L109: const response = await fetch("https://oauth2.googleapis.com/token", { L110: method: "POST", ... L121: if (cachedToken) return cachedToken; L122: if (process.env.GOOGLE_OAUTH_ACCESS_TOKEN) return (cachedToken = process.env.GOOGLE_OAUTH_ACCESS_TOKEN); L123: try { L124: const t = execFileSync("gcloud", ["auth", "print-access-token"], { encoding: "utf8", stdio: ["ignore", "pipe", "ignore"] }).trim(); L125: if (t) return (cachedToken = t);

High

Same File Env Network Execution

A single source file combines environment access, network access, and code or shell execution; review context before blocking.

assets/skills/google-cloud/scripts/gcp.mjsView on unpkg · L108

13// node gcp.mjs audit --project tempo-exla L14: // node gcp.mjs oauth-plan --project tempo-exla --name "Tempo" --home https://tempo.example.com \ L15: // --support you@example.com --redirect https://tempo.example.com/api/auth/callback/google [--redirect …] \ ... L21: L22: import { execFileSync } from "node:child_process"; L23: import { existsSync, readdirSync, readFileSync } from "node:fs"; ... L90: function gcloudConfigDir() { L91: return process.env.CLOUDSDK_CONFIG || join(homedir(), ".config", "gcloud"); L92: } ... L111: headers: { "content-type": "application/x-www-form-urlencoded" }, L112: body: new URLSearchParams({ client_id: cred.client_id, client_secret: cred.client_secret, refresh_token: cred.refresh_token, grant_type: "refresh_token" }), L113: });

High

Command Output Exfiltration

Source combines command execution, command-output handling, and outbound requests; review data flow before blocking.

assets/skills/google-cloud/scripts/gcp.mjsView on unpkg · L13

dist/cloud.jsView file

•Manifest-reachable source captures an API credential, sends it to a fixed unofficial gateway, and persists that redirection in agent or shell configuration. dist/cloud.js: "~/.claude/.credentials.json", "~/.claude/settings.json", "~/.claude/hooks", "~/.claude/CLAUDE.md", "~/.claude.json", "~/.codex/auth.json", "~/.codex/config.toml", "~/.codex/AGENTS.md",

High

Credential Redirect Persistence

Manifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection.

dist/cloud.jsView on unpkg

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/cloud.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/cloud.jsView on unpkg

dist/native/darwin-x64/rudder-nativeView file

•path = dist/native/darwin-x64/rudder-native kind = native_binary sizeBytes = 5395096 magicHex = [redacted]

Medium

Ships Native Binary

Package ships native binary artifacts.

dist/native/darwin-x64/rudder-nativeView on unpkg

dist/improve/collect.jsView file

•stage = ast_semantic_analysis; reason = ast_alias_growth_limit_exceeded; limitedFiles = 7

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/improve/collect.jsView on unpkg

dist/usage.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/usage.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/usage.jsView on unpkg

dist/backends.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/backends.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/backends.jsView on unpkg

dist/board/daemon.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/board/daemon.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/board/daemon.jsView on unpkg

dist/handoff.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/handoff.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/handoff.jsView on unpkg

dist/run-manager.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/run-manager.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/run-manager.jsView on unpkg

dist/task-summary.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/task-summary.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/task-summary.jsView on unpkg

dist/version-check.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/version-check.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/version-check.jsView on unpkg

dist/feedback.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/feedback.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/feedback.jsView on unpkg

dist/improve/state.jsView file

•matchType = normalized_sha256 matchedPackage = @exla/rudder@3.9.6 matchedPath = dist/improve/state.js matchedIdentity = npm:QGV4bGEvcnVkZGVy:3.9.6 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/improve/state.jsView on unpkg

Source & flagged code

24 flagged · loading source

package.jsonView file

•scripts.postinstall = node scripts/postinstall.mjs

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

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

20 High6 Medium5 Low

HighInstall Time Lifecycle Scriptspackage.json

HighChild Processdist/onboarding.js

HighSame File Env Network Executionassets/skills/google-cloud/scripts/gcp.mjs

HighCredential Redirect Persistencedist/cloud.js

HighCommand Output Exfiltrationassets/skills/google-cloud/scripts/gcp.mjs

HighUnverified Remote Native Payload Installpackage.json

HighTrigger Reachable Persistencedist/improve/schedule.js

HighSemantic Analysis Limiteddist/improve/collect.js

HighKnown Malware Source Similaritydist/cloud.js

HighKnown Malware Source Similaritydist/usage.js

HighKnown Malware Source Similaritydist/util.js

Affected versions and remediation

This report applies to @viraatdas/rudder@3.9.7.

See all versions of @viraatdas/rudder for other recorded verdicts.

Review the evidence and your use of @viraatdas/rudder@3.9.7 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: 2026-10-23 00:49:32.505Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Behavioral surface

Source & flagged code