line-editing
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of shell commands to interact with a story management system. This includes the custom 'story' CLI, Node.js for fallback scripts, and system utilities like 'say', 'espeak-ng', and 'spd-say' for audio feedback. While these are for legitimate editing tasks, they represent a significant capability surface.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and analyze untrusted content (prose drafted by users). Maliciously crafted prose could contain instructions intended to override the agent's editing logic or rationale during the 'before/after' proposal process.
- Ingestion points: Markdown files including 'story.md' and various chapter files are read into the agent's context (SKILL.md).
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing external prose.
- Capability inventory: The agent has the ability to execute shell commands, read/write local files (style sheets, character files, chapters), and perform network builds via external engines.
- Sanitization: There is no mention of sanitizing or escaping the content of processed prose before it is used in prompts or CLI commands.
- [DYNAMIC_EXECUTION]: The skill uses dynamic paths to execute a local script ('node ../story-maintenance/scripts/story.js') as a fallback when the primary CLI is unavailable. This creates a dependency on a sibling directory structure that must be trusted.
- [COMMAND_EXECUTION]: The build system allows for user-supplied arguments to be passed to shell commands, specifically the '--pdf-engine' flag. If the agent does not properly validate or sanitize the engine name or path provided by the user, it could lead to command injection.
Audit Metadata