chapter-split
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts (
scripts/split_chapters.py,scripts/generate_nav.py,scripts/merge_multi.py) to process files and usesbun devto launch a documentation preview. These commands are restricted to the local workspace and are necessary for the skill's documented functionality. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill reads external markdown content from
data/markdown/*_pages.mdand configuration data fromstyle-decisions.json. - Boundary markers: Uses
<!-- PAGE N -->markers to identify page boundaries and enforces TOC-based planning to prevent content-based instruction hijacking. - Capability inventory: The skill can write to the
docs/directory and execute project-specific scripts. - Sanitization: Includes a robust validation phase (Step 6) that checks for structural anomalies, such as tables or lists split across file boundaries, which helps verify the integrity of the generated output.
- [SAFE]: No malicious patterns such as obfuscation, credential theft, or unauthorized network activity were identified. The skill's architecture relies on local utilities and structured agent prompts that minimize common attack vectors.
Audit Metadata