ag-update
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches versioning metadata from the author's official GitHub repository at
raw.githubusercontent.com/ag-grid/skills. - [EXTERNAL_DOWNLOADS]: Retrieves upgrade guidelines and changelogs from the vendor's official documentation website at
www.ag-grid.com. - [COMMAND_EXECUTION]: Executes the
npm viewcommand to retrieve the latest version information forag-grid-community,ag-charts-community, andag-studiopackages from the npm registry. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill ingests data from a remote
VERSION.mdfile, official documentation pages, and local project source code during the change identification process defined inreferences/determine-changes.md. - Boundary markers: The skill includes instructions in
SKILL.mdto treat documentation text as verbatim descriptions and specifically forbids the agent from performing specific analysis on the project code using that data, which helps reduce the risk of following instructions embedded in the documentation. - Capability inventory: The skill possesses capabilities to read and write local files (
AG_UPDATE_SCOPE.md,AG_UPDATE_CHANGES.md), execute shell commands (npm view), and invoke sub-agents as detailed inSKILL.mdand its reference files. - Sanitization: It implements logic to filter and classify changes according to version compatibility and product usage, but lacks explicit sanitization or escaping for the external content it interpolates into these files.
Audit Metadata