llvm-security

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill instructs the agent to fetch data from an untrusted source: 'https://raw.githubusercontent.com/gmh5225/awesome-llvm-security/refs/heads/main/README.md'. Under [TRUST-SCOPE-RULE], this is a finding because the user/repo is not in the trusted list.
  • [PROMPT_INJECTION] (MEDIUM): The skill contains an Indirect Prompt Injection surface (Category 8).
  • Ingestion points: External data enters the context from 'https://raw.githubusercontent.com/gmh5225/awesome-llvm-security/refs/heads/main/README.md'.
  • Boundary markers: Absent. The content is directly processed to provide 'detailed information'.
  • Capability inventory: The agent generates security-critical compiler flags, shell commands, and architectural advice.
  • Sanitization: Absent. There is no logic to filter instructions from the data. This allows an attacker who controls the external repository to inject instructions that the agent might follow.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 04:48 AM