test health: findings for azd-copilot

[jongio]

test health

• Overall test coverage is only 30% - the cmd/copilot (main) package and internal/skills package have 0% coverage; cmd/copilot/commands is at 12.1%, internal/copilot at 28.2%, and internal/checkpoint at 34.9%
• The core Launch function in internal/copilot/launcher.go has 0% coverage - this is the primary execution path of the entire CLI and is completely untested
• All checkpoint CRUD operations (List, Get, Latest, Save, SavePhaseCheckpoint, SaveTaskCheckpoint, SaveRecoveryCheckpoint, SaveBeforeDeployment, Delete, Clear, KeepLatest, DetectInterrupted) have 0% coverage despite being a critical feature
• The entire MCP server infrastructure (serveMCP, registerMCPTools, registerMCPResources, and all register*Tools functions in mcp_grpc_tools.go) has 0% coverage
• TestListSkills_Content in skills_test.go uses t.Logf instead of t.Errorf for missing expected skills - the test silently passes even if required skills (azure-deploy, azure-prepare, etc.) disappear from the embedded assets
• TestPlatformSpecificPaths in launcher_test.go is a hollow test - it contains no assertions whatsoever (only t.Log calls), providing zero validation of platform-specific console path handling
• runQuickAction in actions.go has 0% coverage - all five quick-action commands (init, review, fix, optimize, diagnose) share this execution path but none of the actual run logic is tested
• Session runtime handlers listSessions, showSession, and deleteSession in sessions.go all have 0% coverage - command construction is tested but no actual execution paths are exercised

Automated analysis - 8 finding(s)