The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted card data and pipe configurations to diagnose improvements, which could potentially be manipulated by malicious content within cards.
Ingestion points: get_cards in SKILL.md reads up to 50 recent cards, including fields and metadata, providing a surface for adversarial content embedded in card data.
Boundary markers: Absent. There are no instructions to use delimiters or ignore instructions embedded within the card data during analysis.
Capability inventory: The skill possesses write capabilities through create_automation and create_field_condition (defined in SKILL.md), allowing the agent to modify process logic based on its interpretation of ingested data.
Sanitization: Not present. The skill does not explicitly instruct the agent to sanitize or validate card content before processing it for diagnosis.

pipefy-process-intelligence