plot-structure
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill leverages a command-line tool (
story) or a bundled JavaScript fallback (node ../story-maintenance/scripts/story.js) to automate the creation and indexing of story files. The documentation specifically advises usingnoderather thanbunto prevent the accidental execution of local configuration files (likebunfig.toml) that might contain malicious code. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, such as chapter titles, plot beat descriptions, and character names, which are subsequently used as arguments in shell commands (e.g.,
story add chapter '{Title}'). - Ingestion points: User-provided strings for various story elements across multiple instruction files.
- Boundary markers: Absent for shell execution; the skill provides quoting guidance primarily for YAML frontmatter integrity.
- Capability inventory: Execution of shell commands via a local CLI for file system operations.
- Sanitization: The agent is instructed to manually quote values containing special characters (like
#or:) to maintain YAML validity, which offers limited protection against shell-level injection.
Audit Metadata