humanizer-pro
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied AI drafts for humanization, which establishes a potential ingestion point for untrusted instructions. While this creates an indirect prompt injection surface, the risk is negligible due to the skill's restricted capabilities and purely stylistic purpose.\n
- Ingestion points: Processes external "AI drafts" as defined in the Conductor's Protocol in SKILL.md.\n
- Boundary markers: The skill lacks explicit delimiters or instructions for the agent to ignore commands that may be contained within the text to be processed.\n
- Capability inventory: The agent uses internal skill activation and local analysis scripts (scripts/burstiness-audit.py and scripts/inject-eeat-placeholders.ts) with no evidence of network or file-system writing capabilities.\n
- Sanitization: No sanitization or input filtering is mentioned or implemented.\n- [SAFE]: The skill behaves as a standard instructional guide for natural language refinement. It does not perform network calls, access sensitive files, or require credentials. All script references are local and intended for content auditing.
Audit Metadata