Skills
skills/zjp1997720/content-twin-toolkit/writing-clone-starter/Gen Agent Trust Hub

writing-clone-starter

Fail

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSPROMPT_INJECTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Phishing URL detected within the skill's author profile corpus.
  • Evidence: https://wxredian.com/art?id=07bf527cfa52a366210c61375cc30b6f found in references/built-in-profiles/zhouzuoluo/raw-corpus/README.md is flagged by automated scanners as phishing.
  • AV Detection: The file references/built-in-profiles/zhouzuoluo/raw-corpus/README.md has been flagged as suspicious by antivirus software (MD:HttpRequest-inf [Susp]).
  • [OBFUSCATION]: The skill utilizes Base64 encoded URLs to mask external article destinations.
  • Evidence: URLs in references/built-in-profiles/zhouzuoluo/held-out-set.md use the quanlitu.com redirector with Base64 encoded destination parameters (e.g., aHR0cDovL21wLndlaXhpbi5xcS5jb20...).
  • [PROMPT_INJECTION]: Identifies a significant attack surface for indirect prompt injection via untrusted data ingestion.
  • Ingestion Points: The skill instructs the agent to use tools like web-clipper and content-goldmine-gemini to fetch and structured raw articles from the web into the local library (references/profile-distillation/material-sync.md).
  • Capability Inventory: The agent processes this untrusted data to generate profiles and articles, creating a risk that malicious instructions in the source text could override agent behavior.
  • Boundary Markers: There are no explicit instructions or delimiters defined to separate untrusted external content from system instructions during the distillation process.
Recommendations
  • CRITICAL: 1 infected file(s) detected - DO NOT USE
  • AI detected serious security threats
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Apr 28, 2026, 01:27 AM
Security Audit — agent-trust-hub — writing-clone-starter