The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill orchestrates a complex workflow that ingest untrusted user-provided domain descriptions, business requirements, and constraints, which are then carried forward as context for multiple downstream skills.
Ingestion points: User data enters the agent context through responses to the AskUserQuestion tool and initial requirements provided during the interview phase described in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters defined to separate user-supplied requirements from the orchestration instructions, creating a risk that malicious instructions in the requirements could influence the agent's behavior during subsequent workflow steps.
Capability inventory: The skill utilizes the Write tool to store findings in the .trogonai directory and invokes a wide range of external specialized skills (e.g., eventmodeling-brainstorming-events, eventmodeling-elaborating-scenarios).
Sanitization: No evidence was found of input validation, filtering, or escaping of the user-provided data before it is recorded in the project trail or passed to the next stage of the modeling process.

eventmodeling-orchestrating-event-modeling