The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides templates for Azure Data Factory pipelines that ingest untrusted data from external sources and process it using AI services or Large Language Models without explicit sanitization.\n
Ingestion points: Lookup activities in references/ai-services-and-openai-patterns.md and Copy activities in references/databricks-ml-and-e2e-patterns.md read data directly from database tables.\n
Boundary markers: The templates do not demonstrate the use of delimiters or explicit instructions to ignore embedded commands within the processed data.\n
Capability inventory: The skill utilizes WebActivity for REST API calls (Azure AI, OpenAI), DatabricksJob for Spark execution, and SqlServerStoredProcedure for database updates.\n
Sanitization: No explicit sanitization or validation steps are shown for external data before it is interpolated into prompts or sent to AI model endpoints.\n- [DYNAMIC_EXECUTION]: The documentation includes patterns for running dynamic code within SQL environments, including the use of potentially unsafe deserialization.\n
Evidence: references/sql-archival-patterns.md provides an example using sp_execute_external_script which utilizes pickle.loads() to deserialize model data stored in the database.\n
Evidence: references/sql-archival-patterns.md uses dynamic SQL generation with sp_executesql for in-database scoring, although it correctly employs QUOTENAME for basic parameterization.

adf-ml-analytics