gingiris-ugc-matrix
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow that involves ingesting untrusted external data, which creates a vulnerability surface.
- Ingestion points: The 'references/full-sop.md' file (Section XIII: '拆竞品' / Extracting from Competitors) instructs the agent to process external materials: '把收集的素材喂给AI' (Feed collected materials to AI).
- Boundary markers: There are no specific instructions or delimiters provided to ensure the agent ignores potentially malicious prompts embedded within the competitor materials.
- Capability inventory: The skill leverages the agent's core text analysis and processing capabilities to synthesize insights from external data.
- Sanitization: The instructions do not specify any sanitization, filtering, or validation steps for the external content before it is processed by the AI.
- [NO_CODE]: The skill package contains only Markdown files (
README.md,SKILL.md,references/full-sop.md) and no executable scripts, binaries, or automated configuration files that run on the host system.
Audit Metadata