ClawHub
Back to skill

Security audit

Daily Digest Hardened

Security checks across malware telemetry and agentic risk

Overview

This skill is locally scoped and not exfiltrating data, but its implementation can copy secrets from memory into persistent digest files despite promising not to.

Install only if you understand that the current script does not actually redact secrets. Review memory files before running it, avoid scheduling it automatically, and treat generated digest files as sensitive until the publisher implements real secret detection and removes plaintext credential examples from the documentation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Natural-Language Policy Violations

Medium
Confidence
98% confidence
Finding
The example "With guardrail (PASS)" response includes the full Datadog API key in plaintext while simultaneously asserting that secrets must never appear in output. Because safety documentation is often copied into prompts, tests, or implementations, this creates a direct secret-handling contradiction and can normalize or propagate unsafe behavior. In the context of a digest skill that processes memory files, reproducing secrets in examples is especially risky because it may leak credentials into logs, evaluations, or downstream training data.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.