The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill utilizes set_code and with_script methods within references/python-recipe.md to define and execute arbitrary Python logic within the Dataiku platform. This capability allows the agent to generate and run executable code dynamically based on its objectives.
[DYNAMIC_EXECUTION]: The skill uses CreateColumnWithGREL and FilterOnFormula processors, as shown in references/processors.md and references/grel-functions.md, which interpret and execute GREL (General Refine Expression Language) expressions at runtime for data manipulation.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data from external sources (e.g., builder.with_input in SKILL.md and references/prepare-recipe.md) without explicit sanitization. If input datasets contain malicious strings, they could potentially manipulate the logic of transformations or scripts generated by the agent.
Ingestion points: Data is ingested from datasets specified in with_input calls across various reference files (e.g., references/join-recipe.md, references/prepare-recipe.md).
Boundary markers: No boundary markers or 'ignore' instructions are present to prevent the agent from following instructions embedded within the data it processes.
Capability inventory: The skill can execute scripts (recipe.run), write data to storage (write_with_schema), and modify project settings (settings.save).
Sanitization: There is no evidence of input validation or content filtering for the data being processed through recipes.

recipe-patterns