The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted metadata from source databases via schema discovery queries. This metadata (such as table names, column names, and row content) is interpolated into bash commands and SQL scripts without explicit sanitization or boundary markers.
Ingestion points: references/schema-discovery.md (extracts schema metadata like table and column names); references/data-migration.md (extracts table data via CSV/JSON exports).
Boundary markers: The instructions do not define delimiters or "ignore embedded instructions" warnings when processing the source metadata.
Capability inventory: The skill allows the use of the Bash tool for shell command execution and the Write tool for generating migration and validation scripts.
Sanitization: The skill lacks sanitization, escaping, or validation logic for the external identifiers it processes, using simple string placeholders like [table] which could lead to command injection if the metadata is malicious.
[COMMAND_EXECUTION]: The skill provides templates and instructions for executing powerful shell commands (e.g., pg_dump, mysqldump, mongoexport) via the Bash tool. While these are appropriate for the skill's primary purpose of database migration, they represent a high-privilege attack surface if the agent's logic is subverted by untrusted input data.

database-migrator