anti-ai-slop
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including client briefs and existing AI-generated drafts, which serves as a potential vector for indirect prompt injection. While the skill includes robust verification steps (e.g., the 'Hallucination Gate') for factual accuracy, it lacks explicit boundary markers to isolate untrusted input from the agent's core instructions.
- Ingestion points: 'Brief or client source pack' and 'Draft to review' (SKILL.md, references/humanising-rewrite-passes.md).
- Boundary markers: Absent. The skill does not instruct the agent to use specific delimiters or ignore instructions embedded within the provided content.
- Capability inventory: The skill can read provided files, perform web searches for verification (references/humanising-rewrite-passes.md), and generate new content.
- Sanitization: Factual verification passes are present, but there is no mechanism for sanitizing or escaping instructions found within the input data.
Audit Metadata