jacky-write
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided inputs such as voice notes and drafts. It mitigates potential injection or fabrication risks by requiring the agent to build a 'material ledger' to verify facts and by utilizing a 'closed material mode' that restricts the output to only the provided information.
- Ingestion points: Processes user voice notes, ideas, and drafts in SKILL.md and workflow.md.
- Boundary markers: Implements a 'Material Ledger' and 'Closed Material Mode' to ensure the agent adheres strictly to provided facts.
- Capability inventory: The skill is restricted to text generation; it explicitly prohibits automatic publishing or modifying external platform states.
- Sanitization: Includes a 'Platform Safety' gate in references/platform-safety.md to check for forbidden words and commercial risks.
- [EXTERNAL_DOWNLOADS]: The documentation references external GitHub repositories (larashero3-dotcom/lieflat-less-ai-tone and huangbai-AI/guoshen) as sources for writing principles. These resources are credited for influence but are explicitly stated as not being chained or executed at runtime.
Audit Metadata