connect-ai-base
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external enterprise systems (e.g., Salesforce, Jira, SQL Server) via the
queryDatatool, which presents an indirect prompt injection surface. - Ingestion points: Data returned from
queryDataandexecuteProceduretool calls in SKILL.md. - Boundary markers: The skill mitigates risks by requiring explicit discovery steps (
getInstructions,getColumns) and qualified naming ([Catalog].[Schema].[Table]) before data retrieval. - Capability inventory: Includes tool calls for
queryData,executeProcedure, andrequest_upload. - Sanitization: The skill instructs the agent to rely on verified column names and specific SQL dialects rather than guessing based on system output.
- [EXTERNAL_DOWNLOADS]: The skill contains references to official CData Software documentation at
docs.cloud.cdata.comfor SQL references and integration guides. These are trusted vendor resources provided for informational purposes.
Audit Metadata