Changelog¶
Historical release notes below keep version history but use the current developer-focused product language.
Unreleased¶
- narrowed product positioning, prompt-library docs, and marketplace copy to the developer-only coaching scope
- added inline prompt diagnostics and quick-fix code actions to the VS Code companion scaffold
- refactored
docs/prompts.mdinto 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.mdcan 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 software engineers, solution architects, engineering managers, and other tech workflows
- 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.ymland publish-onlyrelease.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
mainand auto-create release commit/tag/notes - enforced release-only version policy by adding
staging-version-guard.ymlfor PRs intostaging - added weekly
stagingtomainsync automation workflow with auto-merge request and admin fallback - added
apply-branch-protection.ymlto 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_TOKENwhen 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 / guardonstaging - updated GitHub Actions dependencies to
actions/setup-python@v6andactions/upload-pages-artifact@v4to stay ahead of the Node 20 deprecation path - enabled GitHub's
FORCE_JAVASCRIPT_ACTIONS_TO_NODE24opt-in across workflows so future Node 24 runner changes are exercised now - fixed Chrome Web Store upload polling to accept both
SUCCESSandSUCCEEDEDupload-state responses from the API
v1.2.15¶
- expanded role-aware Prompt Builder coverage with stronger engineering leadership guidance
- changed profile skill field to level dropdown (
Student,Junior,Middle,Senior) - moved
Studentfrom role mode into level mode for clearer profile semantics - added role-to-template recommendation logic in popup and in-page Prompt Builder
- improved composer detection reliability for ChatGPT/Claude/Grok/Gemini/DeepSeek monitoring and quick builder insertion
- added command shortcut fallback (
Ctrl/Cmd + Shift + O) for opening quick builder when browser blocksCtrl/Cmd + O - improved live bubble resilience with auto-remount behavior on dynamic AI page rerenders
v1.2.14¶
- added role-aware Prompt Builder options by job across core technical coaching flows
- added role selector and role-specific prompt guidance in popup and in-page builder
- changed in-page Prompt Builder shortcut from
F1toCtrl + Owhen composer is focused - made
AI Dev Coach Livebubble always visible on supported AI pages - fixed toast placement consistency to stay top-right on mobile breakpoints
- aligned default overlay duration initialization to 9000ms in background defaults
v1.2.13¶
- added
F1shortcut to open in-page Prompt Builder when AI composer is focused - moved toast notifications to top-right and increased default display duration
- upgraded prompt monitor scoring to evidence-based context checks (error + expected/actual + artifact)
- upgraded independent-attempt detection to quality model (action + result + blocker) with explicit no-attempt penalties
- added bilingual (EN + VI) prompt hint dictionaries for context, attempt quality, and shortcut detection
- added debounced pre-send draft scoring (500ms) for realtime live bubble updates while typing
- fixed live bubble visibility behavior when monitoring is enabled
v1.2.12¶
- fixed prompt capture reliability on ChatGPT by snapshotting composer content before send clears input
- expanded composer selectors for modern editable chat inputs across supported AI platforms
- fixed Live Coach bubble visibility so monitoring HUD remains visible when monitoring toggles are enabled
- improved popup monitoring save flow to auto-enable coach overlays when monitoring is turned on
v1.2.11¶
- improved in-page Prompt Builder launcher visibility with fallback fixed position when chat input is not detected
- launcher now remains available on supported AI pages without requiring extension toolbar click
v1.2.10¶
- added drag-to-reposition behavior for the in-page Live Coach bubble
- persisted bubble position in local storage and restored placement after reload
- clamped drag position to viewport to keep bubble visible on resize
v1.2.9¶
- redesigned popup Live Monitoring into compact switch toggles with immediate auto-save
- improved monitoring section labels for clearer hierarchy and faster scanning
- updated docs home to show latest release tag and author profile summary
v1.2.8¶
- refreshed popup, settings, and in-page coaching UI with pastel design system
- introduced unified primary/secondary palette (
#9DCAEB,#C3EEFA) across all extension surfaces - improved flow with cleaner button grouping, section hints, and elevated visual hierarchy
- added smooth motion for entry/hover states with reduced-motion accessibility fallback
v1.2.7¶
- added in-page Live Coach bubble with realtime prompt score and habit snapshot on AI chat sites
- added prompt-analyzed runtime event bridge so UI updates immediately when user sends a prompt
- persisted last runtime prompt evaluation in storage for stable state after page refresh
- improved send-only tracking path to update live habit stats even when prompt-content reading is disabled
v1.2.6¶
- added in-page Quick Prompt Builder launcher beside AI chat input on supported sites
- added Build + Insert and Build + Send actions to auto-write prompt into chat composer
- hardened auto-send behavior with retry logic and fallback Enter dispatch
- isolated extension-owned UI from prompt/copy/paste monitors to prevent false tracking
- connected quick-builder submit events to prompt analysis pipeline for reliable scoring/stats
v1.2.5¶
- added toggle controls for prompt listener and behavior monitor in popup and settings
- added privacy controls for prompt-content and copied-content reading
- added read-before-copy timing monitor with countdown reminders and fast-copy alerts
- added stats for AI copy events and fast copy behavior
- added bad-prompt and shortcut-prompt counters in habit snapshot
- improved send detection with global Enter/submit fallback listeners
v1.2.4¶
- added immediate prompt analysis on send-button click across supported AI chat UIs
- improved send-time responsiveness by decoupling coach feedback from stats write latency
- added long AI-output copy tracking and warning when pasted back into composer
v1.2.3¶
- added production validator script for extension package integrity
- hardened prompt monitoring and paste detection for better real-world behavior
- improved release packaging and Chrome publish workflow reliability
- expanded Chrome Web Store deployment guide with listing/privacy/assets checklist
v1.2.2¶
- switched docs to MkDocs Material workflow
- aligned docs/nav for extension-focused scope