avoid-eval

Pass

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The file references/rule.md contains an absolute local file system path (/Users/thedaviddias/Projects/frontendchecklist.io/packages/content/rules/en/security/csp.mdx) used as a link target. This reveals the author's internal directory structure and username.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface.
  • Ingestion points: The skill instructs the agent to "Search this codebase" and "Review scripts, client components" within SKILL.md, which involves processing untrusted external source code.
  • Boundary markers: Absent. The skill does not provide delimiters or instructions to treat the ingested code as data rather than instructions.
  • Capability inventory: The agent is tasked with searching the codebase, flagging specific code patterns, and suggesting verification steps, requiring deep analysis of external content.
  • Sanitization: Absent. There are no instructions for the agent to sanitize or escape the content it reviews.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 11, 2026, 12:29 AM
Security Audit — agent-trust-hub — avoid-eval