The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several local scripts to automate prose analysis and formatting checks. These scripts are bundled with the skill and serve the primary purpose of maintaining quality standards.
Scripts include: check-ai-patterns.js (detects repetitive AI writing styles), check-degeneration.js (detects model loops and engineering word leakage), check-delivery-contract.js (verifies word counts and formatting), normalize-punctuation.js (fixes Chinese punctuation), and author_memory_commit.py (manages user writing preferences).
[INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for indirect prompt injection as it processes and analyzes user-supplied story drafts and outlines. However, it mitigates risks by using deterministic regex-based scripts and specialized sub-agents rather than processing raw text directly in a single context without verification.
Ingestion points: 正文.md (Story text), 小节大纲.md (Outline), and 设定.md (Settings).
Boundary markers: Utilizes structured Markdown tables and specific section headers to separate design parameters from creative content.
Capability inventory: Local script execution via node and python3, file writing to the project directory, and tool-based spawning of specialized sub-agents (narrative-writer, character-designer).
Sanitization: Implements extensive prose checking scripts that flag and help remove common AI-generated markers and non-story content.

story-short-write