The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill defines a template that accepts multiple untrusted inputs which are interpolated into the final output without security boundaries.
Ingestion points: The fields completed_stories, sprint_metrics, blockers, and customer_feedback in SKILL.md allow for external data ingestion.
Boundary markers: Absent. The template uses simple {{placeholder}} syntax without instruction to ignore or escape potentially malicious commands within that data.
Capability inventory: No capabilities detected. The skill does not execute subprocesses, write to files, or perform network requests.
Sanitization: Absent. There are no instructions to sanitize or validate the provided sprint data before processing.

sprint-review