ag-update
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches version metadata from a GitHub repository controlled by the author (raw.githubusercontent.com/ag-grid) and retrieves upgrade guides from the official product documentation site (www.ag-grid.com). These sources are recognized as trusted vendor resources.
- [COMMAND_EXECUTION]: The skill executes npm view to query the latest versions of the product packages from the public npm registry. This is a standard method for identifying available updates.
- [PROMPT_INJECTION]: The skill processes external documentation to generate a migration plan, which constitutes an indirect prompt injection surface.
- Ingestion points: Official documentation pages for AG Grid, Charts, and Studio.
- Boundary markers: Not explicitly defined in the output file generation instructions.
- Capability inventory: The agent can execute package manager commands (npm view), write files to the local directory (AG_UPDATE_CHANGES.md), and delete specific local state files (AG_UPDATE_*.md).
- Sanitization: The skill includes logic to strip mentions of external tools (codemods) from the ingested documentation before inclusion in the local plan. The risk is minimized as data is sourced exclusively from the official vendor.
Audit Metadata