The Agent Skills Directory

Command Execution: The skill utilizes local Python scripts for quality assurance and workflow management. It executes scripts/check_economic_impact_report.py to validate report structure, section requirements, and source quality. It also references an internal script user_context_preflight.py for state management. These operations are restricted to the skill's own directory and are part of its standard analytical functionality.
Indirect Prompt Injection Surface: By design, the skill processes external data such as news articles, policy changes, and macro shocks to perform its analysis.
Ingestion points: Untrusted external data enters the workflow through the event definition and research steps documented in SKILL.md and workflow.md.
Boundary markers: The instructions implement protective boundaries by requiring the agent to precisely define the event and clearly separate confirmed facts from inferences or reported scenarios.
Capability inventory: The skill's capabilities include executing local validation scripts and generating complex HTML artifacts.
Sanitization: The skill employs structural validation via the check_economic_impact_report.py script, which enforces a specific report template and checks for source freshness and evidence posture before delivery.

economic-impact-report