constructive-challenge
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted user data, such as project proposals and decision logic, to provide constructive critiques. * Ingestion points: User proposals and discussions as specified in the skill's description. * Boundary markers: None; the skill does not provide specific instructions to the agent on how to handle or ignore potential commands embedded within the user's content. * Capability inventory: No capabilities detected; the skill does not include any allowed-tools, executable scripts, or system-level commands. * Sanitization: None specified; the skill relies on the underlying model's standard input processing.
- [NO_CODE]: The skill is composed exclusively of Markdown documentation. There are no Python scripts, JavaScript files, or configuration files that execute logic or perform actions on the system.
Audit Metadata