The Agent Skills Directory

[SAFE]: The skill instructions are focused entirely on creative writing collaboration, including drafting prose, generating dialogue alternatives, and maintaining character arcs.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests user-provided drafts to match tone and style, while possessing file-writing capabilities to persist output. However, this is the intended primary purpose of the tool and is considered safe within the context of a writing assistant.
Ingestion points: User-provided story drafts and scene fragments provided during the conversation.
Boundary markers: The skill does not explicitly define delimiters for user-provided drafts.
Capability inventory: The skill writes prose and collaboration notes to markdown files (e.g., {project}-collab-{date}.md) within the user's project directory.
Sanitization: No explicit sanitization or instruction-filtering for the user's draft content is defined.
[COMMAND_EXECUTION]: While the skill defines 'Output Persistence' logic, it describes standard file-writing behavior common to AI agents with filesystem access rather than executing arbitrary shell commands or system-level configuration changes.

story-collaborator