tanstack-ai
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses npx and standard package managers (pnpm, npm, yarn, bun) to fetch and install various packages from the npm registry, including @tanstack/ai, @tanstack/intent, and numerous provider-specific adapters. These resources originate from an established vendor and a well-known public registry.
- [COMMAND_EXECUTION]: It instructs the agent to execute shell commands for project environment inspection (ls), version checking (node, npm view), and dependency management (pnpm add, npm install).
- [REMOTE_CODE_EXECUTION]: The skill uses
npx @tanstack/intent@latestto execute remote tooling directly from the npm registry. This tool automates project configuration and the loading of instruction sets (SKILL.md files) for specific packages. - [INDIRECT_PROMPT_INJECTION]: The skill functions as an architectural router that ingests project configuration data (package.json, lockfiles) and dynamically loads instruction sets based on the project's state. This creates an ingestion surface where malicious content within project files or compromised external instructions could influence agent behavior.
- Ingestion points: Project configuration files (package.json, lockfiles), local directory structures (node_modules), and external instruction files (SKILL.md) loaded via the intent tool.
- Boundary markers: No explicit delimiters or warnings for the agent to ignore instructions embedded in the project metadata are provided.
- Capability inventory: Execution of shell commands via npx, package installation, and file system inspection.
- Sanitization: The skill does not describe explicit sanitization or validation mechanisms for the metadata it reads or the external instructions it loads.
Audit Metadata