bypass-winston-ai
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides educational content on writing styles and refers users to the vendor's official tools. No evidence of credential theft, malicious command execution, or data exfiltration was found.
- [EXTERNAL_DOWNLOADS]: The instructions reference the vendor's own MCP package,
@rephrasy/mcp, and their official websiterephrasy.ai. These are documented as vendor-owned resources and do not represent a security risk. - [INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to process and rewrite user-supplied text. While this involves ingesting untrusted data, the operation is limited to stylistic rewriting within the agent's context.
- Ingestion points: User-provided text drafts or AI-generated content provided for humanization.
- Boundary markers: None explicitly defined in the skill instructions.
- Capability inventory: Text analysis and stylistic rewriting.
- Sanitization: Not explicitly implemented; the skill operates as a standard text processing instruction.
Audit Metadata