skills-best-practices

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists of instructional content and documentation. It does not contain executable malicious patterns, obfuscation, or unauthorized data access commands.
  • [DYNAMIC_CONTEXT_INJECTION]: The skill provides thorough documentation on the dynamic context injection footgun (the !command syntax), warning authors against its misuse and explaining how to properly escape or document the syntax to prevent accidental execution at load time.
  • [EXTERNAL_DOWNLOADS]: The skill references the use of uvx --from skills-ref agentskills validate, a tool for validating skill formatting against the Agent Skills specification. This is a developer-focused utility provided by the ecosystem the skill describes.
  • [REMOTE_CODE_EXECUTION]: Mentions of remote code execution (e.g., curl | bash) are included strictly as negative examples within the security and moderation sections to warn developers against using such patterns in their own skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 09:14 PM