The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing shell commands using database utilities such as psql, pg_dump, pg_restore, pg_basebackup, createdb, and dropdb to perform administrative tasks.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where untrusted data from database dumps or query results could influence agent behavior.
Ingestion points: Database content is ingested via pg_restore during restore operations and psql during sanity checks in the SKILL.md file.
Boundary markers: The skill includes instructions to require explicit confirmation before production restores and to use temporary scratch databases for verification, which act as procedural boundaries.
Capability inventory: The skill has the capability to execute shell commands via Bash for database management as defined in its instructions.
Sanitization: No specific automated sanitization or filtering of the database content is described before it is read by the agent.

postgres-ops