The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the standard system utility wc -m to monitor the length of generated analysis output. This is a common and benign operation for character counting.
[DATA_EXPOSURE]: The skill manages a local backup system for the analyzed text. It reads from user-provided file paths and writes to a specific directory (拆文库/) within the project root. These file operations are consistent with the skill's stated purpose of long-term project management and structured analysis.
[INDIRECT_PROMPT_INJECTION]: The skill serves as a surface for indirect prompt injection as it ingests and processes untrusted novel text provided by users.
Ingestion points: Raw novel text files or text pasted directly into the chat.
Boundary markers: None explicitly defined in the instructions to separate novel content from agent instructions during analysis.
Capability inventory: The agent has capabilities to read/write local files and execute basic shell commands (wc).
Sanitization: No specific sanitization or filtering is applied to the ingested text before processing. This is a low-risk architectural characteristic common to text-analysis tools.

story-long-analyze