crossframe-dialogue
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is composed of markdown-based instructions, protocols, and templates for structured dialogue. It lacks any executable code, network capabilities, or sensitive file access mechanisms.
- [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for processing untrusted user input such as reader replies and consultation queries. 1. Ingestion point: reader replies and user queries (SKILL.md). 2. Boundary markers: Informal text-based separation using templates (e.g., templates/default-short-answer.md). 3. Capability inventory: The skill utilizes only text generation and does not have tools for file system modification, network access, or command execution. 4. Sanitization: No explicit content filtering or technical sanitization is mentioned. However, the internal safety protocols (e.g., protocols/consultation-boundary-protocol.md) act as a structural mitigation by requiring the agent to maintain analytical boundaries and refuse harmful personality characterizations.
Audit Metadata