linkedin-post-style
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues or malicious behaviors were identified within the skill files. The instructions focus on style enforcement and quality control.
- [PROMPT_INJECTION]: Surface for indirect prompt injection via user-supplied notes and drafts.
- Ingestion points: User prompts provided to the agent (e.g., notes, drafts) as described in the Process section of SKILL.md.
- Boundary markers: Absent; source material is processed directly without specific delimiters.
- Capability inventory: Capability to invoke rendering tools (md-to-pdf, concept-to-image) and output text to the user.
- Sanitization: Implicitly managed via the 'Hard Blocks' and 'Anti-Pattern' lists which filter the output for AI-like tells and specific keywords.
- [EXTERNAL_DOWNLOADS]: The skill references external URLs for documentation (Wikipedia) and technical examples (Anthropic). These are well-known and trusted sources, and the references are used for educational or bibliographic purposes rather than for downloading executable code.
- [COMMAND_EXECUTION]: While the skill mentions the use of tools like md-to-pdf and ffmpeg, these are described in the context of companion skill invocations or as examples of how the author produced previous content. There are no instructions for the agent to execute unauthorized shell commands.
Audit Metadata