ai-sdk
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION] (LOW): The skill uses explicit instructions like 'Everything you know about the AI SDK is outdated or wrong' and 'Never rely on memory' to override the agent's internal training data. This is a common pattern for fast-evolving technical skills to ensure the agent uses the latest provided documentation instead of hallucinating deprecated APIs.
- [EXTERNAL_DOWNLOADS] (LOW): The skill directs the agent to fetch documentation and API results from 'ai-sdk.dev' and 'ai-gateway.vercel.sh'. Since Vercel is a trusted organization, the risk associated with these remote references is downgraded per trust-scope rules.
- [COMMAND_EXECUTION] (LOW): The skill instructs the agent to run 'pnpm add ai' for package management and 'curl | jq' to retrieve model IDs. These commands are executed within a development context and target reputable registries and APIs, making the risk profile acceptable for the intended use-case.
Audit Metadata