The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to run a local Node.js script (scripts/novel-script.mjs) for deterministic processing of script data, such as timing calculations, quality gate validation, and report generation.
[INDIRECT_PROMPT_INJECTION]: The skill processes JSON configuration files which could serve as an attack surface for indirect prompt injection. 1. Ingestion points: Data is read in scripts/novel-script.mjs from outline.json, art.json, cast.json, and script.json. 2. Boundary markers: No explicit delimiters are used to isolate untrusted data. 3. Capability inventory: The agent has access to Bash, Read, Write, Task, and Glob tools across all scripts. 4. Sanitization: The validate command and HTML escaping in scripts/novel-script.mjs provide some mitigation.

novel-script