chapter-writing
Warn
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands for environment discovery (ls -d) and relies on a custom CLI tool (story) for project maintenance, word counts, and context gathering.\n- [DYNAMIC_EXECUTION]: The skill executes Node.js scripts from paths outside its own directory, specifically ../story-maintenance/scripts/story.js and user-defined checkout paths (/bin/story.js). This dynamic execution of scripts from relative or variable locations allows execution of code outside the skill's controlled scope.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection.\n
- Ingestion points: Reads project files including story.md, character files in characters/, and plot files via the story context command and manual file access.\n
- Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are used when processing this data.\n
- Capability inventory: The skill has capabilities including shell command execution (ls), script execution (node), and file system writes (chapters/, scenes/, continuity/state.md).\n
- Sanitization: There is no evidence of sanitization or filtering of the content ingested from the story project files before it is processed by the agent.\n- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to check for and potentially recommend the installation of an external tool from an unverified GitHub repository (github.com/forjd/better-writing).
Audit Metadata