skills/awslabs/mcp/aws dsql/Gen Agent Trust Hub

aws dsql

Pass

Audited by Gen Agent Trust Hub on Feb 27, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface through the risk of SQL injection. This is due to a limitation in the provided database tools that do not support parameterized queries, requiring the agent to manually interpolate values into SQL strings using regex or allowlists for sanitization. Ingestion points include data retrieved from the database via readonly_query and schema metadata from get_schema, which can influence subsequent commands. The transact tool provides a powerful capability to execute arbitrary DML and DDL, which could be exploited if malicious data is retrieved and then used in a subsequent transaction. No specific boundary markers are defined to isolate data from code structure.
  • [COMMAND_EXECUTION]: The skill includes bash scripts for managing cloud infrastructure and database connections. These are located in the scripts/ directory and include create-cluster.sh, delete-cluster.sh, and psql-connect.sh. These scripts are functional components of the skill authored by awslabs for its primary purpose of DSQL cluster management and connection handling.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 27, 2026, 03:36 PM