mongodb-schema-design
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to work with a MongoDB MCP server and explicitly lists capabilities for high-privilege operations, including destructive actions like
drop-collection,drop-database, anddelete-many. While the skill includes an 'Action Policy' requiring user confirmation for writes, the availability of these tools to the agent creates a high-impact capability surface. - [INDIRECT_PROMPT_INJECTION]: The skill represents an indirect prompt injection surface as it ingests untrusted data from a live database to provide recommendations. If database values or schemas contain malicious instructions, the agent could be influenced to perform unintended actions. Evidence of this attack surface includes ingestion points like
mcp__mongodb__collection-schemaandmcp__mongodb__aggregate(SKILL.md), combined with a lack of documented sanitization for the retrieved data. - [SAFE]: The skill references the official MongoDB MCP server repository (
github.com/mongodb-js/mongodb-mcp-server). As MongoDB is a well-known service and the repository belongs to their official organization, this reference is documented neutrally and does not escalate the security verdict.
Audit Metadata