The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Node.js scripts provided in the package (state-manager.js and prompt-assembler.js) to handle project initialization and state management. These scripts perform routine file operations such as reading and writing JSON data within a user-specified local project directory (defaulting to ~/novels/).
[INDIRECT_PROMPT_INJECTION]: The skill incorporates an indirect prompt injection surface as part of its core memory management functionality. It reads data previously generated by the AI, including chapter summaries and character profiles, to construct subsequent prompts.
Ingestion points: The skill reads historical data from summary.json, characters.json, and foreshadowing.json located in the project directory via the scripts/prompt-assembler.js script.
Boundary markers: There are no explicit boundary markers or delimiters used when interpolating this historical data into the final writing prompt.
Capability inventory: The skill has the capability to execute local scripts and perform file read/write operations in the local filesystem.
Sanitization: The skill does not perform sanitization or schema validation on the content retrieved from its JSON databases before including it in the prompt, relying on the model's inherent ability to follow the new instructions while maintaining the context.

dream-novel-writer