writing-beats
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided markdown files as "raw material" to generate article content. This creates a surface for indirect prompt injection where malicious instructions embedded in the source material could potentially influence the agent's behavior.
- Ingestion points: User-provided markdown files referenced during the "exploit" phase (e.g., in SKILL.md).
- Boundary markers: Not explicitly defined. The agent is instructed to "mine the pile" without specific delimiters separating user data from instruction context.
- Capability inventory: Reading and writing to files on the local disk (e.g.,
SKILL.mdinstructions 3 and 4). - Sanitization: None specified. The skill is intended to paraphrase or quote directly from the raw material.
Audit Metadata