The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation and management of AI automations that process dynamic data, creating an attack surface for indirect prompt injection.
Ingestion points: Untrusted data enters the context via the prompt argument in the create_ai_automation and update_ai_automation tools defined in SKILL.md.
Boundary markers: The instructions specify the use of %{internal_id} tokens for field interpolation; while these delimit data, they do not sanitize the content of the interpolated fields.
Capability inventory: The skill possesses capabilities to create, update, and simulate automated rules (create_automation, update_ai_automation, simulate_automation) that interact with Pipefy's workflow engine.
Sanitization: The skill requires a pre-flight validation step using validate_ai_automation_prompt to check for prompt errors and field overlaps before an automation is created.

pipefy-automations