Harden CommandTree testing, summaries, execution, and SEO

[MelbourneDeveloper]

TLDR

Adds fail-fast CommandTree test execution while broadening command execution, AI summary, coverage, and website SEO hardening across the branch.

Details

What Was Added

• New AI summary orchestration and state modules: src/summaryOrchestration.ts and src/state.ts.
• Automatic Copilot model resolution for background summary generation, with explicit interactive selection preserved for user commands.
• New tree actions for copying relative/full task paths and making files executable from task context menus.
• Drag support for command-tree task rows using text/uri-list, plain text paths, and CommandTree command IDs.
• Dockerfile task discovery and Docker Compose commands that run with an explicit -f file path.
• Pure helper modules for DB lock artifact cleanup, markdown description parsing, task helpers, and task parameter formatting.
• E2E/unit coverage for activation non-blocking behavior, AI model selection, copy path, make executable, Docker execution, script drag, sort order, summary tooltips, undefined-argument no-ops, markdown parsing, task helpers, and DB lock recovery.
• Website audit skill, Codex skill symlinks, sitemap override, taxonomy SEO handling, and expanded website SEO tests.

What Was Changed/Deleted

• make test now runs fail-fast by using Mocha --bail for unit tests and vscode-test --bail for VS Code-hosted tests, while excluding @exclude-ci auth-gated tests from the main test target.
• test-exclude-ci is now an alias for test, keeping existing workflows on the same fail-fast path.
• VS Code-hosted tests now include E2E, provider, and unit output under the coverage run.
• Extension activation no longer blocks on initial discovery; discovery and automatic summaries run in the background after activation.
• AI summaries now choose a concrete model automatically for background runs instead of opening a picker, while manual model selection still prompts and persists the user's choice.
• Command tree sorting now keeps folders before files, sorts labels case-insensitively, and keeps private make/mise tasks muted and sorted after public tasks without inserting divider rows.
• Make target discovery keeps target line numbers for navigation and removes phony/help priority handling from task ordering.
• Task execution delegates parameter formatting to a production helper instead of private duplicated test logic.
• DB lifecycle state moved into centralized app state, with lock artifact operations moved into a testable pure helper.
• Markdown description extraction moved into a parser helper with direct unit coverage.
• Website metadata, sitemap output, taxonomy page indexing, mobile text wrapping, and docs/blog wording were adjusted for SEO and crawl hygiene.
• Dependency lockfiles, package overrides, ESLint config, PR template, agent instructions, and coverage thresholds were updated.

Spec/Doc Changes

• Updated Agents.md, Claude.md, and .github/PULL_REQUEST_TEMPLATE.md.
• Added .claude/skills/website-audit/SKILL.md and .codex/skills/* links for local skill access.
• Updated AI summaries docs/blog copy and mise task blog copy.
• Added sitemap/taxonomy behavior in the website Eleventy config.

Breaking Changes

No extension runtime breaking changes. The intentional workflow change is that make test is now fail-fast and excludes @exclude-ci tests by default; auth-gated Copilot tests remain outside the default CI path.

How Do The Automated Tests Prove It Works?

make ci passed in this session after the branch changes:

• Lint passed through ESLint and cspell.
• TypeScript compile passed.
• Unit tests passed with 118 passing.
• VS Code-hosted tests passed with 372 passing.
• Coverage completed at 93.11% statements/lines, 88.33% branches, and 94.87% functions, then ratcheted coverage-thresholds.json.

Specific tests added or updated in the diff prove the main behaviors: automatic selection picks a concrete model without prompting, activate() returns while initial discovery is still in flight, copy-path E2E tests, Docker execution E2E tests, make-executable E2E tests, script drag E2E tests, sort-order E2E tests, summary tooltip E2E tests, undefined-argument no-op E2E tests, markdown description unit tests, task helper unit tests, and DB lock recovery unit tests.