writing-shape
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from a file ('the pile') to generate content. This creates a surface where malicious instructions embedded in the raw material could influence the agent's behavior during the 'shaping' session.
- Ingestion points: The markdown file containing raw material read at the start of the session (SKILL.md).
- Boundary markers: Absent. The instructions do not define delimiters for the input pile or provide warnings to ignore embedded commands within that content.
- Capability inventory: File system access for reading input files and appending content to article files (SKILL.md).
- Sanitization: None. The content is 'mined' and reworked into the article directly without validation for instruction-like patterns.
Audit Metadata