narrative-nonfiction
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and act upon user-controlled text files, such as book blueprints and session logs, which serves as a potential vector for indirect prompt injection.
- Ingestion points: The agent is instructed in SKILL.md to read book-blueprint.md and files within the sessions/ directory to maintain project state.
- Boundary markers: The instructions do not specify delimiters or explicit 'ignore' warnings to isolate user content from the agent's internal logic during processing.
- Capability inventory: The skill's primary capabilities involve text analysis and file modification (e.g., using str_replace); it does not utilize network operations, shell execution, or sensitive system file access.
- Sanitization: There is no evidence of logic designed to filter or sanitize instructions embedded within the ingested project files before they are read into the context.
Audit Metadata