The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several local scripts using node and python3 to perform text analysis and state management. Specifically, it runs normalize-punctuation.js, check-ai-patterns.js, and check-degeneration.js for prose review, and author_memory_commit.py and tracking_commit.py to maintain project-specific tracking data in the .story/ and 追踪/ directories. These operations are part of the intended functionality for story engineering and author memory management.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it reads and processes external story data which is then passed to multiple sub-agents for analysis.
Ingestion points: Content is gathered from local text files representing chapters, outlines, and character settings as defined in Phase 1 of SKILL.md.
Boundary markers: The instructions do not implement strict delimiters or "ignore previous instructions" markers when interpolating untrusted fiction content into the prompts for sub-agents like story-architect or narrative-writer.
Capability inventory: The skill can execute shell commands via local scripts and spawn additional AI sub-agents through platform-specific tools (subagent_type).
Sanitization: While the skill runs scripts to check for linguistic patterns, it lacks specific filters or sanitization logic for adversarial prompt content within the ingested fiction text.

story-review