The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local command-line interface, renku, to perform project operations, resolve scenes, and manage narrative beats. These operations (e.g., renku screenplay beats create, renku screenplay beats apply) are necessary for the skill's primary function of designing screenplay beats.
[INDIRECT_PROMPT_INJECTION]: The skill acts on data retrieved from external project files and scene contexts via the renku CLI. While this presents a potential surface for indirect injection if project content is untrusted, the skill employs structured data handling (JSON) and strict requirements for system-generated durable IDs, which mitigates risk.
Ingestion points: Data is ingested from commands like renku screenplay beats context and renku screenplay beats show (SKILL.md).
Boundary markers: The skill uses --json flags for structured output, though explicit narrative delimiters are not specified for the creative descriptions.
Capability inventory: The skill can execute shell commands through the renku CLI and coordinate with a media-producer tool (SKILL.md).
Sanitization: Instructions explicitly forbid the agent from inventing IDs and require validation before mutation (SKILL.md, Non-Negotiables).

scene-beat-designer