@luckydraw/cumulus@0.31.65
RLM-based CLI chat wrapper for Claude with external history context management
AI Security Review
scanned 3mo ago · by lpm-firewall-aiNo malicious install-time chain or covert theft was found, so a publish block is not supported. The intentional permission-bypassing agent shell/network surface is a material dangerous capability and merits a warning.
Decision evidence
public snapshot- dist/lib/gateway.js launches Claude with --permission-mode bypassPermissions.
- dist/lib/tools/bash.js exposes bash -c execution to its agent loop.
- dist/lib/tools/web-fetch.js accepts runtime URL fetches; web-search contacts api.search.brave.com.
- dist/gateway/setup.js can install login services, but only after explicit setup and TTY confirmation.
- dist/lib/version-check.js can run npm install -g only through explicit update commands.
- package.json has no preinstall, install, or postinstall hook.
- package.json prepare only invokes husky; prepublishOnly is publisher-side build/type-check.
- Gateway setup and service persistence are reached only by cumulus-gateway setup.
- Gateway/API operations require explicit runtime start and API-key authentication.
- No source evidence of credential harvesting, covert exfiltration, obfuscated payload loading, or foreign AI-agent config mutation.
Source & flagged code
7 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@0.31.65.
See all versions of @luckydraw/cumulus for other recorded verdicts.
- Review the evidence and your use of @luckydraw/cumulus@0.31.65 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.