explain-diff
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute a local Python script (
render.py) to process JSON specifications and generate HTML explanation pages. - [EXTERNAL_DOWNLOADS]: The generated HTML documentation references the Highlight.js library and associated CSS themes hosted on Cloudflare's CDN (
cdnjs.cloudflare.com). This is documented as a neutral reference to a well-known service. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, specifically code diffs and pull request content, to generate summaries and quizzes.
- Ingestion points: Code changes, branches, and pull request data provided as input for documentation generation.
- Boundary markers: The skill relies on structured sections (Background, Intuition, Code, Quiz) but does not define explicit delimiters to isolate the diff content from agent instructions.
- Capability inventory: Local script execution (
render.py), file system write operations for outputting HTML files, and network capability via external MCP tools (Notion/Outline). - Sanitization: The
render.pyscript performs HTML escaping for text and code blocks, and the metadata format specifies character escaping to prevent arbitrary HTML execution within the Markdown comments.
Audit Metadata