webmcp-implementation
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides comprehensive architectural guidance for implementing WebMCP with a strong emphasis on security boundaries. It explicitly instructs the agent to reuse existing authorization checks, validate all inputs, and maintain human confirmation for sensitive actions.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of a tool-calling interface in a target application, which introduces a surface area for indirect prompt injection. The skill mitigates this risk by providing specific instructions for implementing boundary markers (such as untrusted content hints), strict schema validation, and data minimization strategies.
- [SAFE]: The diagnostic script (
scripts/probe-webmcp.js) is a read-only tool designed for environment verification. Technical analysis confirms it does not perform tool registration, network operations, or unauthorized data access. The accompanying test suite (scripts/probe-webmcp.test.cjs) uses standard Node.js testing frameworks to verify the script against local synthetic environments. - [SAFE]: All external references target trusted documentation sources, including Chrome's developer portal, W3C community drafts, and official agent platform documentation.
Audit Metadata