The Agent Skills Directory

[SAFE]: The skill performs standard narrative orchestration tasks using a predefined workflow. It does not attempt to exfiltrate data, execute unauthorized commands, or bypass security constraints.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data as part of its primary function to generate narrative content.
Ingestion points: User-supplied narrative content description and repository-local files including AGENTS.md, CLAUDE.md, and production/review-mode.txt.
Boundary markers: The skill relies on standard platform tool boundaries but does not implement specific semantic delimiters or "ignore previous instructions" guards for external content.
Capability inventory: Includes file system access (Read, Write, Edit), repository searching (Glob, Grep), and sub-agent orchestration via the Task tool.
Sanitization: There is no evidence of explicit sanitization or validation of the ingested narrative data before it is passed to sub-agents.

team-narrative