package-upgrade
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources such as npm changelogs and GitHub releases, as well as local package manifests. This presents a surface for indirect prompt injection if malicious instructions are embedded in package documentation. However, the skill is strictly scoped to analysis and reporting, includes a mandatory human-in-the-loop approval gate (Phase 5), and does not have the capability to execute the code it analyzes.
- [EXTERNAL_DOWNLOADS]: The skill performs web research on official platforms like npmjs.com and GitHub to retrieve package versions and migration guides. These references are directed toward established, well-known services for software development tasks and are used for informational purposes only.
- [SAFE]: No malicious command execution, data exfiltration, or obfuscation patterns were detected. The skill maintains a clear workflow focused on inventory and assessment within a restricted report directory (
tmp/reports/).
Audit Metadata