The Agent Skills Directory

COMMAND_EXECUTION (MEDIUM): The orchestrator script scripts/reverse-outline.ts uses Deno.Command to execute various analysis scripts. This requires the --allow-run flag, a high-privilege permission that allows the agent to execute arbitrary system commands, although the script specifically targets local files in the scripts/ directory.
EXTERNAL_DOWNLOADS (LOW): The orchestrator script dynamically imports standard modules from https://deno.land/std@0.208.0/. These are external dependencies retrieved at runtime, which is standard for Deno but represents an external link.
PROMPT_INJECTION (LOW): The skill is vulnerable to indirect prompt injection because it is designed to ingest and analyze untrusted text files (e.g., book.txt).
Ingestion points: The script scripts/reverse-outline.ts reads the inputFile provided by the user.
Boundary markers: Absent. The skill does not implement delimiters or 'ignore instructions' markers when passing book content to analysis tools.
Capability inventory: The orchestration environment has allow-run, allow-read, and allow-write permissions across the analysis pipeline.
Sanitization: While the inferBookName function sanitizes the book's title for safe filesystem use, the actual text content of the book is not sanitized or escaped before processing.

reverse-outliner