system-design-validator
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: No critical or high-severity security threats were detected. The skill does not attempt credential theft, data exfiltration, or unauthorized network communication.
- [COMMAND_EXECUTION]: The skill uses a local script,
scripts/validate_design.py, to perform deterministic validation of system design files. This script is part of the skill package and performs standard JSON processing and rule-based logic, representing a safe use of local execution for its intended purpose. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) because it ingests and processes external design documents that could contain malicious instructions.
- Ingestion points: The skill reads
design.jsonfiles and reference documents provided in the agent's context. - Boundary markers: The skill instructions do not specify the use of delimiters or clear separation between system instructions and untrusted data content.
- Capability inventory: The skill has the capability to execute local validation scripts and generate structured architectural findings.
- Sanitization: While the Python script validates the JSON schema and specific fields, there is no sanitization of the natural language text within the design fields to prevent it from being interpreted as instructions by the LLM.
Audit Metadata