beyond-code-think
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions are limited to design and architectural clarification. It explicitly prohibits implementation code modification and uses no high-privilege tools or dangerous commands.- [INDIRECT_PROMPT_INJECTION]: The skill processes user requirements to generate specification files, creating a minor surface for indirect instructions through user intent.
- Ingestion points: User-provided architectural inputs and clarification responses used to populate Stage 3 specification templates.
- Boundary markers: Uses a fixed Markdown structure for
spec.mdand explicitly defined stages to separate processing steps. - Capability inventory: The tool is permitted to read and write files solely within the
.beyond-code/subdirectory. - Sanitization: No specific output sanitization is mentioned, but the instructions strictly limit the agent to writing non-executable documentation.
Audit Metadata