ai-sdk-model-manager
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (LOW): The skill performs
npm installfor packages under the@ai-sdkorganization. These are managed by Vercel, which is a recognized trusted organization. Per [TRUST-SCOPE-RULE], the risk is downgraded to LOW. - [COMMAND_EXECUTION] (LOW): The skill executes multiple shell commands, including project-specific test suites (
npm run test) and GitHub CLI operations (gh pr create). While standard for developers, these involve executing scripts that could be modified in the local environment. - [PROMPT_INJECTION] (LOW): The use of a 'researcher subagent' to pull data from external websites creates an Indirect Prompt Injection surface (Category 8).
- Ingestion points: External web content fetched during Step 3.
- Boundary markers: None explicitly specified for the research context.
- Capability inventory: File writing,
npm install, andnpm run testexecution. - Sanitization: No explicit sanitization or filtering of web content before it is proposed for inclusion in the codebase.
- Mitigation: The skill incorporates a mandatory user approval step (Step 4) and quality checks (Step 8) which act as a human-in-the-loop firewall.
Audit Metadata