inspect

Audit Claude Code plugins by running a local scan of manifests, hooks, scripts, and SKILL.md files to detect prompt injection, data exfiltration, and credential access. The audit runs entirely offline and produces a structured report with severity ratings and remediation guidance.

Security risks in Claude Code hooks and SKILL.md files include prompt injection, data exfiltration, unauthorized credential access, and obfuscated code. Scanning plugin manifests and configurations identifies these permission abuse vectors and generates actionable PASS, WARN, or FAIL findings.

Yes, you can perform a local security audit of installed plugins entirely offline. The audit evaluates plugin manifests, hooks, scripts, and SKILL.md files locally to detect malicious behavior and produces a structured security report without requiring network connectivity.

Detect prompt injection and data exfiltration in plugin scripts by scanning for obfuscated code and suspicious configurations within the local plugin ecosystem. The audit process identifies these threats and reports them with severity levels and specific remediation steps.

The best way to review Claude Code plugin permissions for abuse vectors is to read plugin manifests and configurations through an automated local audit. This process identifies potential permission abuses and generates a structured report with PASS, WARN, or FAIL findings to guide remediation.

Yes, the plugin security audit provides remediation guidance for detected vulnerabilities. The structured report assigns PASS, WARN, or FAIL findings with severity levels and actionable steps to resolve identified prompt injection, data exfiltration, and credential access risks.