The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documentation details a Retrieval-Augmented Generation (RAG) pattern which involves processing untrusted document chunks and user queries.
Ingestion points: Data enters the system via the chunks table content column and user question embeddings processed in SKILL.md.
Boundary markers: The skill describes retrieving context to be fed into an LLM, though specific prompt-level delimiters are not explicitly defined in the provided snippets.
Capability inventory: The skill utilizes database interaction capabilities via the mariadb Python connector.
Sanitization: Code examples strictly adhere to security best practices by using parameterized SQL queries (cur.execute(..., (params,))) to prevent SQL injection attacks.
[EXTERNAL_DOWNLOADS]: The skill references official MariaDB packages (mariadb, langchain-mariadb) and documentation from mariadb.org and mariadb.com. These are recognized as official resources from the skill's authoring vendor.
[COMMAND_EXECUTION]: Provides standard SQL syntax and CLI usage examples for interacting with MariaDB's native vector functions, which is expected for this type of technical guide.

mariadb-vector