novel-indexing-and-searching

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documentation provides commands to execute local Python scripts (build_index.py and search.py) for maintaining project metadata. This is standard behavior for the tool's purpose and operates entirely within the local workspace.
  • [DYNAMIC_EXECUTION]: The build_index.py script utilizes ast.literal_eval to parse metadata from file frontmatter. This is a safe method for evaluating literal structures and is used here for internal configuration parsing without exposing the system to arbitrary code execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection as it processes text from novel files to build indexes.
  • Ingestion points: Markdown files located in the novel/ workspace directory (e.g., bible, draft, continuity folders).
  • Boundary markers: None identified in the parsing logic.
  • Capability inventory: The scripts perform local file system read and write operations within the designated workspace directory.
  • Sanitization: No explicit sanitization or filtering of the novel content is performed before indexing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 03:32 PM