Security audit
Security checks for vulnerabilities and agentic risk
This package provides Claude access through Google Vertex AI, with credential use and network activity consistent with that purpose.
The provider can automatically detect existing Google application-default credentials. Use credentials with appropriate project permissions, confirm your region and billing limits, and use only trusted endpoint overrides. Model requests send conversation content to the configured service. SkillSpector was inapplicable because the package contains no bundled skills; the packaged runtime was inspected directly.
SkillSpector was not run because this plugin release contains no bundled skills.
Detected: suspicious.env_credential_access
function createAnthropicVertexStreamFn(projectId, region, baseURL, deps = defaultAnthropicVertexStreamDeps, env = process.env) {