build-webmcp-tools
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references several development and testing tools hosted on the NPM registry, including
use-webmcp-tool,webmcp-types,webmcp-evals, andchrome-devtools-mcp. These are vendor-provided or well-known utility packages intended for the development and debugging of WebMCP applications. - [COMMAND_EXECUTION]: The instructions guide developers to execute shell commands such as
npx webmcp-evalsfor verifying tool schemas andnpx chrome-devtools-mcpfor site discovery. These operations are legitimate parts of the documented developer workflow. - [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface for indirect prompt injection because it processes user-supplied application logic, role-play transcripts, and tool descriptions. However, it incorporates significant defensive measures, including a dedicated security guide (
agent-security.md) that explains the threat model and provides specific mitigation strategies such as the use ofuntrustedContentHintand structural spotlighting.
Audit Metadata