@luckydraw/cumulus@1.0.67
RLM-based CLI chat wrapper for Claude with external history context management
AI Security Review
scanned 15d ago · by lpm-firewall-aiSource inspection found package-aligned CLI, skill, and update behavior, plus a postinstall cleanup of unused GPU provider files. The reviewed credential copy is for a restricted local Codex run; no malicious chain was established.
Decision evidence
public snapshot- The postinstall hook runs a GPU-provider cleanup script; it is not an agent-configuration installer.
- The gateway agent-related location searches standard paths for the Claude executable.
- The restricted Codex path copies authentication into a scratch directory for an isolated local CLI run.
- Skill files are stored in Cumulus's own skills library, and installation is an explicit function call.
- The Claude process hook writes its prepared prompt to the subprocess stdin.
- The gateway setup source reads its config and redacts key, secret, and token values in its display path.
- The version updater installs a selected Cumulus release globally when invoked.
- The gateway daemon exposes explicit start, stop, and status commands.
Source & flagged code
9 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgPackage source references child process execution.
dist/tui/hooks/useClaudeProcess.jsView on unpkg · L3Findings
Affected versions and remediation
This report applies to @luckydraw/cumulus@1.0.67.
See all versions of @luckydraw/cumulus for other recorded verdicts.
- Review the evidence and your use of @luckydraw/cumulus@1.0.67 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.