refactored docs/prompts.md into a richer prompt library with structured multi-line templates and reusable placeholders
updated prompt parsing, previews, and marketplace rendering so longer prompts stay readable in popup cards and inline suggestions
added inline prompt suggestions inside supported AI chat inputs, using local prompt-library matching, role-aware ranking, and popularity boosts
added a dedicated prompt suggestion engine and smoke coverage for inline suggestion ranking
added Prompt Marketplace with local markdown-backed prompt browsing, search, filtering, and trending
added marketplace prompt actions to copy, insert, or send prompts into supported AI chat pages
added prompt-library sync tooling so docs/prompts.md can drive the extension bundle
refreshed popup prompt output with a structured preview card, metadata badge, and clearer raw-text fallback
redesigned live coaching bubble into a smaller stats-first layout with readable prompt preview and coaching signals
polished in-page Quick Prompt Builder with grouped sections, better helper copy, and clearer action feedback
added a shared role-coaching module so popup builder, quick builder, live monitoring, and analytics can use the same role-aware hints and warnings
added role-specific coaching examples and warnings for Teacher, Software Engineer, Solution Architecture, Manager, Director, Doctor, and Other profiles
added software-engineering specialization guidance for frontend, backend, DevOps, and fullstack wording
improved generated prompt formatting with cleaner section breaks for easier reading and copy/paste
added a lightweight local trend dashboard with quality-over-time, warning-frequency, and prompt-category views
added daily session summaries to learning analytics with prompt category breakdowns and friendly coaching suggestions
added local learning analytics event tracking for prompt submissions with popup snapshot summary
added future backend sync payload contract for prompt-session analytics rollout
added learning-analytics smoke checks in CI
added prompt linter with rule-based checks for short prompts, missing technical context, missing failure signals, and sensitive data
added prompt-linter smoke checks in CI and popup lint results UI
added prompt quality engine v2 with shared scoring between popup prompt builder and live monitoring
added prompt-quality smoke checks in CI to guard scoring regressions
replaced direct-to-main release bumping with protected-branch-safe release PR automation
split release CI into prepare-release.yml and publish-only release.yml
added manual CI dispatch support so release PR branches receive required checks
updated workflow actions to current major versions for checkout and setup-node
added local secret detection and redaction guardrail before prompt submission
added sprint planning documentation with prioritized backlog estimates
split CI/CD so docs deploy and extension release run independently
constrained docs deployment trigger to docs-related changes only
rebuilt extension release pipeline to bump manifest version only
added automatic friendly release-note generator for non-technical audiences
redesigned release workflow to auto-bump semantic version on main and auto-create release commit/tag/notes
enforced release-only version policy by adding staging-version-guard.yml for PRs into staging
added weekly staging to main sync automation workflow with auto-merge request and admin fallback
added apply-branch-protection.yml to manage branch protection policy as code
fixed release tagging in GitHub Actions by configuring bot git identity before annotated tags are created
hardened release preparation and weekly sync to use REPO_ADMIN_TOKEN when repository policy blocks PR creation from the default Actions token
changed Chrome Web Store "item in review" handling so GitHub release workflows finish successfully and report store publish as deferred
aligned branch protection automation with merge-commit release flow and required Staging Version Guard / guard on staging
updated GitHub Actions dependencies to actions/setup-python@v6 and actions/upload-pages-artifact@v4 to stay ahead of the Node 20 deprecation path
enabled GitHub's FORCE_JAVASCRIPT_ACTIONS_TO_NODE24 opt-in across workflows so future Node 24 runner changes are exercised now
fixed Chrome Web Store upload polling to accept both SUCCESS and SUCCEEDED upload-state responses from the API