@reason-machines/cli-win32-arm64@0.2.416
⚠ Under reviewReason CLI native binary for win32-arm64
AI Security Review
scanned 4d ago · by lpm-firewall-aiThe manifest declares no install hook, and the inspected JavaScript excerpt does not establish an attack. The primary native executable contains unresolved behavior: its bounded analysis shows credential-file-reading code candidates but cannot establish reachability or exclude other behavior, so manual review is warranted.
Decision evidence
public snapshot- The manifest identifies this as a Windows ARM64 native CLI package; it declares no lifecycle hooks.
- The flagged runtime contains code selecting .claude/commands and .claude/agents paths, but this excerpt alone does not establish a write or its activation.
- The host analysis found embedded code that reads local credential files, but its bounded windows do not establish whether that code is reachable or whether credentials are sent.
- The 86 MB executable’s analysis explicitly cannot establish loader reachability or rule out other native behavior; the ARM64 native module was unreadable as source.
- No install lifecycle hook is declared in the inspected package manifest.
- The inspected excerpts do not establish credential transmission or another concrete attack.
Source & flagged code
6 flagged · loading sourcePackage source references child process execution.
bin/runtime/srt/sandbox-runtime.mjsView on unpkg · L60A single source file combines environment access, network access, and code or shell execution; review context before blocking.
Findings
Affected versions and remediation
This report applies to @reason-machines/cli-win32-arm64@0.2.416.
See all versions of @reason-machines/cli-win32-arm64 for other recorded verdicts.
- Review the evidence and your use of @reason-machines/cli-win32-arm64@0.2.416 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.