The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains a Node.js script verify-book.mjs that executes a sibling script verify-doc.mjs using spawnSync. This is part of the intended verification workflow to delegate specific checks.
[DYNAMIC_EXECUTION]: The verification script is capable of creating and writing JSON files to the local file system (e.g., figures/book-map.scene.json) when the --write flag is provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided markdown and JSON configuration files to perform linting and verification. While this presents a theoretical attack surface for indirect prompt injection, the script logic is focused on static analysis and uses safe execution methods that do not invoke a shell.
Ingestion points: Reads the contents of book.json, checks.json, and all markdown chapter files (e.g., 01-quickstart.md) within the target directory.
Boundary markers: The script does not use explicit boundary markers to separate user data from agent instructions during its internal processing, though it uses regular expressions to isolate specific sections.
Capability inventory: The script has capabilities for file system access (readFileSync, writeFileSync, mkdirSync) and subprocess execution (spawnSync).
Sanitization: The script uses standard JSON parsing and regex-based content extraction. The spawnSync command is executed with an argument array and without shell invocation, preventing command injection via malicious filenames.

explainer-book