ai-core
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill documents the use of an MCP (Model Context Protocol) stdio transport in
tool-calling/SKILL.md, which allows applications to launch local processes (e.g., usingnpx) to serve as tool backends. This is a standard feature for MCP integration. - [INDIRECT_PROMPT_INJECTION]: The skill naturally processes untrusted data from external sources such as LLM responses and tool outputs. This represents a known attack surface for indirect prompt injection; however, the library utilizes strict schema validation (via Zod, ArkType, or Valibot) as a primary mechanism to sanitize and structure these inputs.
- [EXTERNAL_DOWNLOADS]: The skill includes instructions for installing official TanStack companion packages and utilities (such as
@tanstack/intent) via standard package managers, which is normal practice for this ecosystem.
Audit Metadata