The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes performance data from external sources (such as ad platform exports or API data) and interpolates this content into prompts for content generation and media buying analysis. This creates a surface for instructions embedded in the advertising data to influence agent behavior.
Ingestion points: Performance data for active creatives, including metrics, names, and IDs sourced from external ad platforms (MCPs) or user-provided exports.
Boundary markers: The skill instructions do not specify any delimiters or warnings to the model to ignore embedded instructions within the performance data.
Capability inventory: The skill executes a Python script (creative-fatigue-predictor.py), writes files to the local system (reports and briefs), and coordinates multiple specialized agents for content creation and media buying.
Sanitization: No explicit sanitization, validation, or escaping of the ingested marketing data is described in the workflow.
[DYNAMIC_EXECUTION]: The skill is designed to run a specific Python script, creative-fatigue-predictor.py, which is part of the skill's package. This script performs the primary fatigue scoring calculations on the provided performance data.
[COMMAND_EXECUTION]: The processing workflow involves invoking the creative-fatigue-predictor.py script as a subprocess, passing it performance data to generate risk levels and health scores.

creative-health