The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes database metadata that could contain malicious instructions designed to influence the agent.
Ingestion points: The exploration workflow in references/EXPLORATION_PLAYBOOK.md fetches comment fields from duckdb_tables(), duckdb_views(), and duckdb_columns(). The skill also previews data rows.
Boundary markers: The skill does not provide delimiters or system instructions to help the agent distinguish between data/metadata and its own operational instructions.
Capability inventory: The skill uses SQL tools (query, query_rw) to interact with the database environment.
Sanitization: There is no evidence of metadata validation or sanitization before it is presented to the agent.
[COMMAND_EXECUTION]: The skill provides templates for executing SQL commands on the target data platform.
Evidence: references/EXPLORATION_PLAYBOOK.md contains multiple SQL code snippets for discovering schemas, listing databases, and profiling tables.
[EXTERNAL_DOWNLOADS]: The skill connects to remote database endpoints and attaches shared data sources via URL.
Evidence: The TypeScript starter connects to pg.us-east-1-aws.motherduck.com. The playbook describes using ATTACH '<share_url>' to connect to remote data shares. These are standard vendor operations.

motherduck-explore