The Agent Skills Directory

[SAFE]: The skill interacts exclusively with the local system through the renku CLI. All network references target the vendor's official domain at gorenku.com. No unauthorized external connections or data exfiltration patterns were detected.
[SAFE]: File system interactions are strictly scoped to the project's tmp/ and previs/source/ directories. The skill explicitly forbids direct modification of core database files like .renku/project.sqlite, ensuring project integrity.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a transformer for screenplay data into shot plans, which constitutes an ingestion surface for indirect prompt injection. However, this risk is mitigated by the mandatory use of structured JSON contracts (shot-plan-json-contract.md) and focused CLI commands that validate data before mutation, effectively sanitizing the input pipeline.
[SAFE]: No obfuscation, persistence mechanisms, or privilege escalation attempts were found. The skill maintains a clear separation of concerns, delegating media generation to the media-producer agent rather than executing generative tasks itself.

shot-planner