test: improve cmd coverage from 63.4% to 67.6%

[JordanCoin]

Summary

• Add focused table-driven tests for cmd/skill.go behaviors (RunSkill usage, list/show output, init idempotency)
• Add focused table-driven tests for cmd/serve.go JSON/error response helpers
• Extend cmd/drift_test.go coverage for resolveCodePaths glob-cleanup/fallback behavior
• Raise CI coverage floor from 50.0 to 55.0 in .github/workflows/ci.yml

Coverage

Scope	Before	After
cmd/	63.4%	67.6%
TOTAL	72.7%	73.9%

Validation

• go test -race -coverprofile=coverage.out ./...
• go vet ./...
• gofmt -l .
• No source files modified, only _test.go files plus CI coverage floor update
• Coverage floor bumped in CI (50.0 -> 55.0)

[Copilot]

Pull request overview

This PR improves test coverage for the cmd/ package by adding new table-driven tests around key CLI behaviors and HTTP response helpers, and tightens CI by raising the enforced coverage floor.

Changes:

• Add new tests for cmd/skill.go subcommand behavior (usage output, list/show output, init idempotency).
• Add new tests for cmd/serve.go JSON and error response helpers.
• Extend cmd/drift_test.go to cover resolveCodePaths glob cleanup and fallback behavior; raise CI coverage floor to 55%.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
cmd/skill_test.go	Adds stdout-based tests for skill CLI usage/list/show/init behaviors.
cmd/serve_test.go	Adds tests validating JSON/error response helpers (headers, status codes, JSON validity).
cmd/drift_test.go	Adds coverage for resolveCodePaths configured-path cleanup and guessed-path fallback.
.github/workflows/ci.yml	Raises enforced coverage floor from 50.0% to 55.0%.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

These tests invoke runSkillList/runSkillShow which call skills.LoadSkills(root). LoadSkills also reads global skills from the user’s home directory (~/.codemap/skills), so local developer environments with custom/overriding skills can change the output and make these assertions flaky (e.g., no remaining "[builtin]" lines or "Source: builtin"). Consider making the tests hermetic by setting HOME (and on Windows USERPROFILE) to t.TempDir() (or otherwise disabling global skill discovery) before running the commands.

[Copilot]

The field name wantPrefix is a bit misleading here: the assertion checks for an exact match (p == tt.wantPrefix) rather than a prefix. Consider renaming it to something like wantPath/wantEntry (or update the assertion to actually check a prefix) to keep the test intent clear.