writing-beats
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill focuses on structured text generation and local file operations (reading raw material and writing an article file) consistent with its stated purpose as a writing assistant.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external content from a markdown file ('raw material') and re-reads the 'article file' from disk, which are potential ingestion points for untrusted data.
- Ingestion points: Raw material markdown file and user-edited article file (SKILL.md).
- Boundary markers: None explicitly defined in the instructions for the ingested data.
- Capability inventory: File read and file write (append-only beat generation). No network or command execution capabilities.
- Sanitization: None specified for the input data.
- While instructions could be hidden in the raw material to influence the agent's output, the skill lacks dangerous capabilities that would allow for significant harm, and the interaction is under human supervision (choose-your-own-adventure style).
Audit Metadata