data-integrity-tester
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists exclusively of documentation and templates (Markdown and CSV files) intended for human or AI agent use in planning software testing. It does not include any scripts, binaries, or automated execution logic.
- [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to ingest external project requirements and architecture documents for analysis. While this creates a data ingestion surface, the skill lacks any automated capabilities—such as shell execution, network access, or file modification tools—that could be exploited via malicious instructions embedded in those documents. The skill's output is limited to text-based reports and design matrices.
- Ingestion points: Project-level instructions, architecture, and requirements documents (SKILL.md, line 11).
- Boundary markers: None present for external data inputs.
- Capability inventory: None. The skill contains no executable scripts or system commands.
- Sanitization: None. The skill assumes the agent will process provided text for documentation purposes.
Audit Metadata