rainskills@0.1.28
Interactive Rainbond skill installer for Codex, Claude Code, and Pi Agent
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThe package has no install lifecycle hook or self-dependency, so the evidence does not support a malware block. Its explicit CLI can still modify AI-agent control surfaces and remote SSH authorization, so it warrants a warning under the dangerous-capability policy.
Decision evidence
public snapshot- The user-run installer writes skills into Codex, Claude, and Pi agent directories and replaces differing skill trees.
- A normal runtime-status command can fetch a newer package release and replace installed skill directories.
- The explicit SSH preparation command generates a key and appends its public key to the selected server's authorized keys.
- The installer posts lifecycle telemetry to a Rainbond endpoint.
- The installer includes instructions addressed to AI assistants.
- package.json has no preinstall, install, or postinstall hook.
- Agent and SSH changes are reached through explicit CLI commands, not npm installation.
- The updater accepts only fixed registries, validates SHA-512 integrity, and runs npm with scripts disabled.
- No source evidence shows credential harvesting, secret exfiltration, hidden shell execution, or a runtime self-dependency.
Source & flagged code
5 flaggedThe user-run installer writes skills into Codex, Claude, and Pi agent directories and replaces differing skill trees.
install.shView on unpkg · L1053Findings
Affected versions and remediation
This report applies to rainskills@0.1.28.
See all versions of rainskills for other recorded verdicts.
- Review the evidence and your use of rainskills@0.1.28 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.