The Agent Skills Directory

[PROMPT_INJECTION]: Potential for Indirect Prompt Injection.
Ingestion points: The skill retrieves metadata and query details from a Trino database through tools such as list_tables and get_query_detail in SKILL.md.
Boundary markers: The instructions do not specify markers to delimit data or warn the agent to ignore instructions embedded in the retrieved data.
Capability inventory: The agent can execute SQL commands via execute_query and explain_query.
Sanitization: No sanitization logic is provided in the instructions for the ingested data.
[COMMAND_EXECUTION]: Database Query Operations.
The skill allows execution of SQL commands on remote Trino instances. The instructions define a workflow limited to 'inspection-only' queries like EXPLAIN and SHOW, advising against DML/DDL use.

mcp-trino