cloudbase-document-database-web-sdk
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data from CloudBase NoSQL collections, which serves as a potential vector for indirect prompt injection if the stored data contains malicious instructions targeting the LLM.
- Ingestion points: Data retrieved through
db.collection().get()(crud-operations.md), aggregation results (aggregation.md), and realtime snapshots via.watch()(realtime.md). - Boundary markers: The skill does not explicitly instruct the agent to use delimiters or 'ignore instructions' warnings when processing data retrieved from the database.
- Capability inventory: The skill enables full CRUD capabilities, geolocation queries, and permission management via
managePermissions, providing a significant set of tools that could be misused if the agent is compromised by injected data. - Sanitization: While the skill provides examples of data validation logic (e.g., the
validateTodofunction incrud-operations.md), this is primarily for schema/data integrity rather than prompt injection mitigation.
Audit Metadata