database-architect
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
Bashtool to implement a 'Memory Protocol'. This involves reading and writing to local files within the.claude/context/memory/directory (learnings.md, issues.md, decisions.md) to persist state and configuration between sessions. - [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface (Category 8). 1. Ingestion points: User-provided database requirements, entities, and access patterns described in
SKILL.md. 2. Boundary markers: Not explicitly defined for user input interpolation. 3. Capability inventory: IncludesBash,Write, andEdittools for managing schema files and migrations. 4. Sanitization: No explicit sanitization or validation of the input requirements is described in the provided files. While this represents a standard risk surface for database engineering agents, no malicious intent was identified.
Audit Metadata