@luckydraw/cumulus@1.0.4
RLM-based CLI chat wrapper for Claude with external history context management
AI Security Review
scanned 3mo ago · by lpm-firewall-aiThe package is not malicious by the install-time or covert-behavior boundary, but it ships a high-risk, broadly privileged agent execution surface with limited default containment.
Decision evidence
public snapshot- dist/lib/gateway.js spawns Claude with --permission-mode bypassPermissions.
- dist/lib/tools/{bash,write,web-fetch}.js expose unrestricted shell, filesystem writes, and arbitrary HTTP to model tool calls.
- dist/lib/gateway.js embeds operational AI instructions including local authenticated agent-message posting.
- dist/gateway/setup.js can install a persistent LaunchAgent/systemd service after an explicit setup prompt.
- package.json has no preinstall/install/postinstall hook; prepare only invokes husky.
- Persistence is reached only through the explicit cumulus-gateway setup command and interactive consent.
- Network code targets configured LLM providers, npm update checks, or user-supplied WebFetch URLs; no hard-coded exfiltration endpoint found.
- No source evidence of hidden credential harvesting, covert payload download, or automatic destructive behavior.
Source & flagged code
8 flagged · loading sourcePackage source references child process execution.
dist/tui/hooks/useClaudeProcess.jsView on unpkg · L3This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/lib/gateway.jsView on unpkgFindings
Affected versions and remediation
This report applies to @luckydraw/cumulus@1.0.4.
See all versions of @luckydraw/cumulus for other recorded verdicts.
- Review the evidence and your use of @luckydraw/cumulus@1.0.4 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.