sharpen
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a textual prompt refinement tool that does not perform any network operations, file system modifications, or external script execution. The agent configuration restricts implicit tool usage, ensuring a limited execution environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied tasks, which presents a surface for indirect prompt injection.\n
- Ingestion points: The skill ingests a "supplied task" as the primary data for its rewriting process in
SKILL.md.\n - Boundary markers: The skill lacks explicit boundary delimiters (such as XML tags or triple backticks) to separate user input from its own instructions.\n
- Capability inventory: No tools, network calls, or dangerous subprocesses are defined or utilized by the skill.\n
- Sanitization: While no sanitization is performed, the skill explicitly instructs the agent to treat input as text for rewriting and to stop before execution, mitigating the risk of following instructions embedded within the user data.
Audit Metadata