@su-record/vibe@4.1.22
An AX/FDE harness for Claude Code, Codex CLI and ChatGPT desktop. Say what you need, approve the scenarios once; the harness proves the work by running the checks itself. Memory in plain files, human tokens for irreversible actions, a ledger instead of cl
AI Security Review
scanned 11d ago · by lpm-firewall-aiStatic inspection confirms automatic, opt-out guarded setup of this package’s agent surfaces during global installation. This is a first-party agent extension lifecycle risk, not evidence of a confirmed malicious attack.
Decision evidence
public snapshot- The package postinstall runs a setup script that proceeds on global installs unless VIBE_SKIP_SETUP is set.
- The setup script calls setupGlobal, which installs surfaces for each detected client home.
- Home setup writes an instruction card, copies skills, and adds a notification hook to client configuration.
- The plugin installer assembles a local plugin tree and updates a marketplace entry.
- The register helper reads Claude plugin registration data; this excerpt does not show credential handling or a remote payload.
- The install hook exits for non-global installs and supports an explicit VIBE_SKIP_SETUP opt-out.
- The inspected install behavior sets up this package’s own agent surfaces; the cited code shows no unrelated agent-control mutation or credential exfiltration.
Source & flagged code
15 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgPackage source references child process execution.
mcpb/server/index.jsView on unpkg · L10Findings
Affected versions and remediation
This report applies to @su-record/vibe@4.1.22.
See all versions of @su-record/vibe for other recorded verdicts.
- Review the evidence and your use of @su-record/vibe@4.1.22 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.