The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected. The skill follows security best practices by including explicit prohibitions against storing secrets or PII in design documents and requiring manual approval before any database modifications (DDL) are performed.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface for indirect prompt injection as it is designed to process untrusted external data, such as database requirements, sample CSV/JSON data, and existing schema files. However, the risk is mitigated by strong instructional boundaries that prioritize manual review and require explicit user approval for any downstream actions. * Ingestion points: Requirements analysis in SKILL.md and data inspection workflows in references/workflow.md involving user-provided files. * Boundary markers: Present. The skill mandates 'Do not execute DDL by default' and 'Present DDL and migration steps for approval.' * Capability inventory: Use of inspection tools (vd:miudb, sqlit) and the dbdocs CLI for validation. * Sanitization: Relies on the agent's logic to model data rather than executing it, with a requirement for human-in-the-loop approval for any schema changes.

dbdesign