The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references documentation and source code from official and trusted sources, including YugabyteDB, PostgreSQL, SQLAlchemy, OpenAI, and the LangChain-AI GitHub organization.
[COMMAND_EXECUTION]: Code examples demonstrate database interactions using standard libraries like psycopg and langchain-postgres. All SQL examples use parameterization (e.g., cur.execute(sql, (params))) to prevent SQL injection attacks.
[CREDENTIALS_UNSAFE]: The skill correctly advises using environment variables (os.environ["YSQL_CONNINFO"]) and TLS for production deployments. While it includes a connection string with default credentials, this is explicitly labeled as a "local-development example" for localhost involving the vendor's own product default setup.
[INDIRECT_PROMPT_INJECTION]: As a RAG-focused skill, it handles external document ingestion. The instructions mitigate injection risks by advocating for typed predicates and parameterized SQL, and explicitly warning against substituting raw user input into SQL commands.

yb-rag-langchain