author-profile
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill manages a workflow where user-provided writing preferences are stored and subsequently used to influence agent decisions, which is a standard injection surface.\n
- Ingestion points: User-provided author preferences, writing goals, and style requirements as described in SKILL.md.\n
- Boundary markers: The instructions explicitly define precedence rules (e.g., current instructions always override stored preferences) to mitigate potential logic bypasses.\n
- Capability inventory: Basic file write and read operations for
workspace/AUTHOR_PROFILE.mdand coordination with memory management functions.\n - Sanitization: No explicit sanitization or filtering of stored user content is described.\n- [SAFE]: The skill does not contain any executable code, remote script downloads, or network exfiltration patterns. It focuses entirely on instructional logic for local file management.
Audit Metadata