You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The service package (service.go) has 0% test coverage despite being the core business logic layer specifically designed with dependency injection for testability - BuildRequestOptions, Execute, NewRequestService, DefaultTokenProviderFactory, and DefaultHTTPClientFactory are all completely untested.
The config package has 0% test coverage - config.Defaults() and the Config struct are never exercised in tests, leaving default values and config construction unverified.
TestClient_Execute_RetryExponentialBackoff is a potentially flaky test - it uses hard absolute timing assertions (interval1 >= 50ms, interval2 > interval1, interval3 > interval2) with no tolerance band, making it susceptible to failures under CPU contention, slow CI runners, or OS scheduling jitter.
The formatResponse error path in mcp.go (75% coverage) - the branch where json.MarshalIndent fails is untested, leaving the MCP fallback error serialization unverified.
NewMCPCommand's RunE (the actual MCP server startup via server.ServeStdio) is never exercised in tests (66.7% coverage), so any initialization or startup error in that path is invisible to the test suite.
validateScopeURLMatch has 86.7% coverage - the single-label scope host rejection branch (e.g., a bare TLD like "com" as scope) lacks a dedicated test, leaving a security-relevant guard uncovered.
test health
servicepackage (service.go) has 0% test coverage despite being the core business logic layer specifically designed with dependency injection for testability -BuildRequestOptions,Execute,NewRequestService,DefaultTokenProviderFactory, andDefaultHTTPClientFactoryare all completely untested.configpackage has 0% test coverage -config.Defaults()and theConfigstruct are never exercised in tests, leaving default values and config construction unverified.TestClient_Execute_RetryExponentialBackoffis a potentially flaky test - it uses hard absolute timing assertions (interval1 >= 50ms,interval2 > interval1,interval3 > interval2) with no tolerance band, making it susceptible to failures under CPU contention, slow CI runners, or OS scheduling jitter.formatResponseerror path in mcp.go (75% coverage) - the branch wherejson.MarshalIndentfails is untested, leaving the MCP fallback error serialization unverified.NewMCPCommand'sRunE(the actual MCP server startup viaserver.ServeStdio) is never exercised in tests (66.7% coverage), so any initialization or startup error in that path is invisible to the test suite.validateScopeURLMatchhas 86.7% coverage - the single-label scope host rejection branch (e.g., a bare TLD like"com"as scope) lacks a dedicated test, leaving a security-relevant guard uncovered.Automated analysis - 6 finding(s)