docs: streamline agent guidance and maintain relevant skills

[erseco]

Reduce AGENTS.md to current repository constraints and commands, preserving WordPress 6.1/PHP 8.0 compatibility, archive/security boundaries, translation rules, and the distinction between release and source archives. Move PHPUnit coverage attribution detail into a task-specific reference.

Install official WordPress block development, performance and project-triage skills, plus GitHub Actions hardening and Microsoft Playwright CLI, using gh skills. Reinstall the five previously untracked upstream skills with source metadata so they can be refreshed; this also updates the security-audit skill and its validators. Keep upstream text verbatim, preserve Claude symlinks, and include license texts.

Add a Monday/manual update workflow with version tags for all skill-update actions, job-scoped write permissions, a main base, and skill-only PR changes. Updates require review and never auto-merge.

Local validation: lint, PHPUnit (943 tests / 2074 assertions), translation checks and architecture validation passed. PHPMD passed with an existing installed binary against this repository's ruleset; the make target itself could not find a global phpmd executable. Skill validator tests, frontmatter/link/provenance checks, actionlint and diff checks passed.

The update command was exercised in dry-run mode and repository workflow permissions permit PR creation. Scheduled/manual workflow execution after merge has not been exercised. Default GITHUB_TOKEN update PRs do not start CI automatically; review their prompt changes. No unrelated architecture decision or release is introduced. Upstream trailing blank lines are preserved verbatim; the staged whitespace check ignores only blank-at-EOF warnings.

Prepared with assistance from OpenAI Codex, informed by:

• OpenAI: Rethinking skills and prompts for GPT-6 Astra
• Anthropic: Prompting Claude Opus 5
• Anthropic: Effective context engineering for AI agents
• Anthropic: Equipping agents for the real world with Agent Skills

Applied principles: task-specific discovery, concise repository constraints, progressive disclosure of domain details, and preserving authorization boundaries without repetitive approval rituals. Anthropic documentation informed the guidance; this does not claim a separate Claude review.

Action versions: actions/checkout@v7 and devantler-tech/actions/update-agent-skills@v13.3.3. Upstream currently has no v13 tag.

peter-evans/create-pull-request@v8; no skill-update action remains SHA-pinned.

[github-actions]

Test in WordPress Playground

Test the plugin with the code from this branch:

ℹ️ The eXeLearning editor is fetched from the shared release and unpacked into the plugin when the playground boots, so the first load may take a few extra seconds. ELP upload, shortcode, Gutenberg block and preview work normally.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.87%. Comparing base (fe88c50) to head (736a788).

Additional details and impacted files

@@            Coverage Diff            @@
##               main     #123   +/-   ##
=========================================
  Coverage     96.87%   96.87%           
  Complexity      864      864           
=========================================
  Files            39       39           
  Lines          4323     4323           
=========================================
  Hits           4188     4188           
  Misses          135      135           

Flag	Coverage Δ
javascript	95.70% <ø> (ø)
php	97.25% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.