story-maintenance

Pass

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a Node.js script to perform filesystem operations such as reading, writing, renaming, and deleting markdown files within a story project. The script implements robust safety measures, including path resolution checks (assertSafeProjectPath, assertLexicallyInsideRoot) and symbolic link rejection (rejectSymlinkTarget), to ensure operations remain confined to the project root and prevent directory traversal.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from markdown files.
  • Ingestion points: Project markdown files (chapters, character profiles, story bible) and external files processed during the import command.
  • Boundary markers: Absent. The instructions do not define delimiters or provide warnings to the agent regarding instructions embedded within the story content.
  • Capability inventory: Filesystem write (fs.writeFileSync), delete (fs.rmSync), and rename (fs.renameSync) capabilities within the project root (located in scripts/story.js).
  • Sanitization: The script uses structured YAML parsing for metadata and regular expression-based patterns for prose analysis, which provides some isolation between the metadata and the prose content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 12, 2026, 08:22 PM
Security Audit — agent-trust-hub — story-maintenance