The Agent Skills Directory

Prompt Injection (SAFE): No instructions found that attempt to bypass safety filters or override system-level instructions.
Data Exposure & Exfiltration (SAFE): No hardcoded secrets or sensitive file paths detected. The skill specifically instructs the user to filter by shopId to prevent multi-tenant data leaks, which is a defensive security measure.
Obfuscation (SAFE): The content is entirely human-readable Markdown and code. No hidden characters, Base64 encoding, or homoglyphs are present.
Unverifiable Dependencies & Remote Code Execution (SAFE): The skill mentions npx prisma commands for migrations, which is standard tooling for the stated purpose. No untrusted remote scripts or shell pipelining were found.
Indirect Prompt Injection (SAFE):
Ingestion points: The skill describes patterns for reading database records (products, shops) and parsing JSON metadata.
Boundary markers: The skill recommends using explicit Prisma where clauses and select statements to define clear data boundaries.
Capability inventory: The skill is granted Read, Write, Edit, Grep, and Glob tools, which are appropriate for managing database-related code files.
Sanitization: Error handling for common database exceptions is demonstrated, though specific sanitization of untrusted database content for the LLM context is out of scope for this documentation-style skill.
Persistence Mechanisms (SAFE): No attempts to modify system startup files, cron jobs, or shell profiles were identified.

database-best-practices