database-design

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill's documentation, covering database selection, indexing, and migration strategies, contains only legitimate technical guidance without malicious instructions or prompt injection attempts.- [SAFE]: The scripts/schema_validator.py script is a standard local utility. It uses built-in Python libraries to perform structural checks on Prisma and Drizzle schema files. It does not exhibit risky behaviors such as network connectivity, sensitive file access, or dynamic code execution.- [INDIRECT_PROMPT_INJECTION]: The skill includes a schema validator that processes external project files, which constitutes a potential ingestion surface for untrusted data. 1. Ingestion points: scripts/schema_validator.py reads content from file paths identified via glob. 2. Boundary markers: Absent. The script prints extracted identifiers and error messages directly. 3. Capability inventory: The skill environment allows Read, Write, Edit, Glob, and Grep operations. 4. Sanitization: Absent. Substrings from input files are interpolated into printed issue messages. This surface is assessed as safe given the tool's intended use-case for static code analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 07:20 AM
Security Audit — agent-trust-hub — database-design