Agent Audit Trail

• Outcome: Pass
• Latest audit: May 28, 2026 at 11:36 AM UTC
• Version:
  2.1.0

Overview

This is a coherent local audit logging template, but the resulting audit log may contain sensitive operational metadata and should be protected.

Before installing, decide where the audit file will live, who can read it, how long it will be retained, and what fields must be redacted. Do not put raw secrets, tokens, passwords, unnecessary personal data, or overly detailed credential locations in audit entries; treat any future remote log shipping as a separate security and compliance decision.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill explicitly recommends logging credential access events and rich provenance details, including examples like `.env:ACTP_KEY_PASSWORD` and session identifiers, but it does not warn users that the audit log itself can become a sensitive data store. In a compliance-focused logging skill, this increases the chance that secrets, secret locations, user identifiers, or operational metadata are copied into a broadly accessible append-only log and retained longer than intended.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The roadmap proposes remote log shipping to S3/GCS but does not warn that audit data may contain sensitive operational, identity, provenance, or security-relevant records that would be transmitted to third-party cloud providers. Because this skill is designed for comprehensive event logging, remote export materially expands the exposure boundary and can create confidentiality, residency, and compliance risks if enabled without safeguards.

VirusTotal

65/65 vendors flagged this skill as clean.