The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to generate and execute SQL commands against MotherDuck databases. It includes defensive instructions for handling high-risk operations like ATTACH, DETACH, and SHUTDOWN, mandating the use of specific write-capable tools (e.g., query_rw) only upon explicit user confirmation.
[DATA_EXPOSURE]: The skill facilitates data analysis and exploration. While it enables access to database content, it follows best practices for credential management (requiring a prior connection via motherduck-connect) and does not contain hardcoded secrets or unauthorized data exfiltration patterns.
[INDIRECT_PROMPT_INJECTION]: The skill has a potential surface for indirect prompt injection as it processes and presents results from external databases.
Ingestion points: SQL query results returned from MotherDuck databases (referenced in SKILL.md).
Boundary markers: Absent; the instructions do not specify delimiters for database output.
Capability inventory: Execution of SQL queries (including read and write capabilities) and analytical transformations.
Sanitization: Absent; the skill relies on standard agent handling of structured SQL results.
[REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of untrusted scripts or the installation of unverified packages. It explicitly warns against installing extensions at runtime as a common mistake.

motherduck-query