triflux@10.71.0
CLI-first multi-model orchestrator for Claude Code — route tasks to Codex, Antigravity, and Claude
AI Security Review
scanned 9d ago · by lpm-firewall-aiThe postinstall automatically installs package-owned agent skills, which is a guarded agent extension lifecycle risk. The inspected behavior does not establish credential theft, an unrelated network transfer, or another concrete attack, so this warrants a warning rather than a block.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review
Trigger
Installing the package through npm, which invokes its postinstall script.
Impact
Package-provided skills may affect the user’s Claude agent behavior after installation; no malicious effect is established by the inspected source.
Mechanism
The setup flow calls skill synchronization, which copies package skill files into the Claude skills directory. Codex hook approval is separately deferred or prompted for consent.
AI rationale
The postinstall automatically installs package-owned agent skills, which is a guarded agent extension lifecycle risk. The inspected behavior does not establish credential theft, an unrelated network transfer, or another concrete attack, so this warrants a warning rather than a block.
Evidence
package.jsonscripts/setup.mjs
Decision evidence
public snapshotAI called this Suspicious at 83.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The package postinstall runs its setup script automatically.
- Setup invokes skill synchronization during its setup flow.
- That synchronization targets the user’s Claude skills directory and copies package skill files there.
- Codex hook approval is deferred in noninteractive runs and otherwise requires a user prompt.
Evidence against
- The inspected setup behavior is limited to package-provided skills and setup changes; it does not establish a concrete attack.
- The inspected source provides no evidence of credential collection or an unrelated network recipient.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsUrlStrings
Source & flagged code
25 flagged · loading sourcepackage.jsonView file
•scripts.postinstall = node scripts/setup.mjs
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkgbin/tfx-live.mjsView file
1#!/usr/bin/env node
L2: import { execFile } from "node:child_process";
L3: import { randomBytes } from "node:crypto";
High
bin/triflux.mjsView file
1530});
L1531: // ── psmux 기본 셸 자동 수정 (cmd.exe → PowerShell) ──
L1532: if (process.platform === "win32" && which("psmux")) {
High
Shell
Package source references shell execution.
Findings
1 Critical20 High6 Medium5 Low
CriticalAi Agent Control Hijackscripts/setup.mjs
HighInstall Time Lifecycle Scriptspackage.json
HighChild Processbin/tfx-live.mjs
HighShellbin/triflux.mjs
HighExternal Ai Agent Control Surface Mutationscripts/mcp-check.mjs
HighTrigger Reachable External Ai Agent Control Surface Mutationscripts/ensure-codex-hooks.mjs
HighTrigger Reachable Persistencescripts/setup.mjs
HighRuntime Package Installbin/triflux.mjs
HighSemantic Analysis Limitedhub/team/headless.mjs#virtual:normalized:round1
HighKnown Malware Source Similarityscripts/lib/mcp-health.mjs
HighKnown Malware Source Similarity
Affected versions and remediation
This report applies to triflux@10.71.0.
See all versions of triflux for other recorded verdicts.
- Review the evidence and your use of triflux@10.71.0 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.