Product Updates

Changelog

All the latest updates, improvements, and fixes to Git AutoReview.

Coming SoonQ1 2026

Fixed

Telemetry stops reporting failures that never happened — Usage events were being delivered and stored correctly, but the confirmation arrived a fraction of a second too late, so the extension reported an error and retried every minute. The output panel stays quiet now

Coming Soon

Planned

Custom AI Prompts — Fine-tune AI behavior for your codebase
Review Templates — Reusable templates for common review scenarios

v1.21.9Latest

August 8, 2026

Fixed

Clear message when your Claude usage limit is reached — Agent reviews run on your Claude subscription, and once its usage window is exhausted requests are queued until it resets. Git AutoReview now says so right away, including when the limit resets, instead of leaving you waiting and reporting a generic timeout

v1.21.8

August 8, 2026

Fixed

Reviews keep their code context — A code index interrupted mid-write could leave every later review running with no project context at all, quietly lowering the quality of results. The index is now written safely, repairs itself if it was damaged, and stays lean in large monorepos

Improved

Changed

Faster agent reviews — Claude CLI reviews finish sooner and are far less likely to run out of time before returning results. Prefer depth over speed? Set Claude CLI Effort to high in Settings
Smaller download — The extension package is about 1.3 MB lighter on every platform

v1.21.7

August 8, 2026

Improved

Changed

Sharper code context in monorepos — Build caches and generated output (Nx, Turborepo, Next.js, and similar) are no longer treated as source code, so reviews draw on your real files instead of build artifacts. Your code index refreshes automatically after this update

v1.21.6

August 6, 2026

Fixed

Reliable reviews on the newest Claude models — Reviews with Claude Opus 5, Sonnet 5, and Fable 5 now always return their full results (previously they could come back empty) and have more room for detailed findings
Accurate model info — GPT-5.6 Luna now shows its true 1M-token context window in the model picker
Fewer background requests — Shared team key lookups make fewer network round-trips during reviews and settings changes

Improved

Changed

Thorough preset tuned — The Thorough preset now pairs Claude Opus 5 with Gemini 3.5 Flash, Google's strongest stable model, for deeper multi-model reviews
More resilient releases — Publishing to the Open VSX registry now surfaces real failures instead of hiding them, so every platform build reliably reaches both marketplaces

v1.21.5

August 5, 2026

New

Added

Latest AI models — The model picker now includes the newest releases across every provider: Claude Opus 5, Sonnet 5, and Fable 5 (Anthropic), Gemini 3.6 Flash (Google), and GPT-5.6 in its Sol, Terra, and Luna variants (OpenAI). Choose a top-tier flagship for the deepest reviews, or a fast, cost-effective model for quick feedback

Improved

Changed

Fresher defaults — New installations now default to Gemini 3.6 Flash and GPT-5.6 Terra where those providers are used — newer, faster, and more cost-effective. Your existing model choices are unchanged (if you had a retired Gemini preview model selected, reviews switch to Gemini 3.6 Flash automatically)

Fixed

Shared team keys now work across multiple teams — If you belong to more than one team, Git AutoReview now finds and uses the shared API key wherever your team manager set it, so reviews no longer fail with "API key not configured"

v1.21.4

July 4, 2026

Improved

Changed

Marketplace listing details — Maintenance release refreshing the extension's listing metadata. No functional changes

v1.21.3

July 3, 2026

Improved

Changed

Improved reliability and reduced background network usage — Git AutoReview now backs off automatically when our servers are briefly unreachable, then recovers on its own once they're back — using less background network in the meantime. Everyday reviewing is unchanged and no action is needed on your part

v1.21.2

June 7, 2026

Fixed

Correct default model on fresh installs — New installations now start with Claude Sonnet 4.6 by default, as intended — a balanced 1M-token model. Existing setups are unaffected

v1.21.1

June 7, 2026

New

Added

Claude Opus 4.8 — Anthropic's newest flagship model is now available in the model picker. Select it for the most capable reviews; the Thorough preset uses it automatically. The balanced default is unchanged

v1.21.0

May 27, 2026

New

Added

Latest AI models — The model picker now includes the newest releases across every provider: Claude Opus 4.7, GPT-5.5 and GPT-5.4, and Gemini 3.5 Flash — plus refreshed options for Claude CLI and Copilot CLI. Choose a top-tier flagship for complex reviews, or a fast model for quick feedback

Improved

Changed

Smarter default model — New installations now default to Claude Sonnet 4.6, a 1M-token model that balances review depth with cost. Change it anytime in Settings

Fixed

Only models you can actually use — Removed AI models that the providers have discontinued, so choosing a model never leads to a failed review

v1.20.1

May 24, 2026

Improved

Changed

Research mode quality — Refinements to the optional invitation-only research participation mode. No effect on regular reviewing

v1.20.0

May 21, 2026

New

Added

Research participation mode (invitation-only) — If you receive a personal invitation link from an academic study using Git AutoReview, the link now opens the study directly in a dedicated window. Ignore this entirely if you have not been invited — it has no effect on your everyday reviewing

v1.19.3

April 17, 2026

New

Added

Multi-language code understanding — Reviews for 11 languages (JavaScript, TypeScript, Python, Java, Go, Rust, C#, Ruby, PHP, C/C++, Swift) now use symbol-level analysis instead of pattern matching. Polyglot codebases get sharper context and more relevant AI suggestions

Fixed

Comments land on the right line — Review findings now point to the exact code being discussed, every time. Previously, comments could appear on nearby lines when the PR version differed from your local copy

v1.19.2

April 9, 2026

New

Added

Visual PR context — Screenshots and images embedded in PR descriptions are now sent to the AI alongside your code diff. Claude, GPT, and Gemini see the UI changes you're reviewing — not just the code. Works with GitHub, GitLab, and Bitbucket

Improved

Changed

Smarter Jira ticket matching — The extension now prioritizes the Jira key in the PR title over the branch name, so you always get the right ticket context when they differ
More reliable rate limit handling — GitHub secondary rate limits (HTTP 403) are now retried automatically with proper backoff, instead of failing immediately
Safer concurrent reviews — Switching to a different PR while a review is running no longer shows stale results from the previous PR

Fixed

Event buffer reliability — Fixed a race condition where telemetry events queued during a network flush could be silently lost
Quality scores for large PRs — Quality Assessment scores now appear correctly for PRs large enough to be reviewed in chunks

v1.19.1

April 8, 2026

Fixed

Review progress overlay — The loading overlay no longer allows scrolling the content behind it during agent reviews

v1.19.0

April 8, 2026

Improved

Changed

Faster, more reliable reviews — Optimized how reviews are processed across GitHub, GitLab, and Bitbucket. Large PRs with many comments now load completely instead of stopping at the first page
GitHub Enterprise support — Pagination now works correctly for GitHub Enterprise Server instances
Stronger security — Tightened content security policies across all extension panels. AI-generated content is now sandboxed to prevent any possibility of code execution
Better resource management — The extension now cleans up background processes, timers, and listeners more thoroughly, keeping VS Code fast even during long sessions
Smarter multi-model reviews — When using two AI models together, each model's findings are now correctly attributed. The "choose best suggestion" view shows the actual model names instead of generic labels
More accurate reviews — Improved how the extension handles edge cases in GitLab merge requests, especially for older self-managed instances
Agent review diagnostics — Added detailed logging for Claude CLI agent reviews, making it easier to troubleshoot when reviews take longer than expected

Fixed

Review comments now load completely — Previously, PRs with more than 100 comments could silently miss older discussions. All platform comments are now fetched reliably
Bitbucket Server pagination — Fixed an issue where comment threads could be incomplete on Bitbucket Data Center instances with high activity
Approve and reject buttons — Fixed a rare case where approving or rejecting a comment could affect the wrong suggestion
Edit Comment panel — Icons now display correctly in the comment editing view
Extension reload — Fixed a memory leak that could occur when reloading the extension window repeatedly
OpenAI key validation — API keys are now validated more accurately, preventing false matches with similarly-formatted keys from other providers

v1.18.10

April 2, 2026

Added

Shift-Left Review — Review staged changes before you commit. Click the review button in Source Control or run "Review Staged Changes" from the Command Palette — catch bugs and security issues before they reach a pull request

Fixed

Team sync — Team members now appear in the team panel automatically after activating their license key

v1.18.9

March 25, 2026

Fixed

Team repository limits — Team members no longer hit a false repository limit after reopening Settings. The unlimited quota now persists correctly across sessions

v1.18.8

March 23, 2026

Added

Custom Endpoints — connect any LLM — Run code reviews with Ollama, LM Studio, Groq, Together AI, or any OpenAI-compatible model. Add your endpoint in Settings, test the connection, and it joins your review alongside Claude, Gemini, and GPT
Test Connection — One-click connection test before saving your custom endpoint — no guessing if the URL or model ID is right

Fixed

Team members can now add repositories — Members on a Team plan could see their repos but couldn't add new ones. The repository limit now correctly reflects the team's unlimited quota

v1.18.7

March 18, 2026

Added

File Review Rules — You decide what gets reviewed. New Settings section lets you define which files always get a full AI review and which are skipped as trivial — per team, per project
Copy comments — One-click copy button on every review finding and inline comment — paste directly into your PR discussions
Git blame context — AI agent reviews now understand code history — who wrote each line and when — for smarter, more relevant feedback

Fixed

CI/CD workflow reviews — GitHub Actions, GitLab CI, Bitbucket Pipelines, Dockerfiles, and Jenkinsfiles now get full AI review by default

v1.18.5

March 14, 2026

Fixed

Platform comments in diff view — Inline comments from GitHub, Bitbucket, and GitLab now appear directly in the diff editor where they were left
Comment authors — PR comment authors now display correctly across all platforms
PR switching — Platform comments are properly cleared when switching between PRs

v1.18.4

March 14, 2026

Added

Platform comments — See existing GitHub, Bitbucket, and GitLab comments directly in the review panel and as inline code threads

Improved

Changed

Logging — Color-coded, level-filtered output in the VS Code Output panel for easier debugging

Fixed

Bitbucket credential warning — No longer shown when only GitHub or GitLab repositories are configured

v1.18.3

March 13, 2026

Fixed

Removed debug command — Internal testing command was visible in command palette

v1.18.2

March 13, 2026

Fixed

Add Repository default — GitHub is now the default platform when adding a repository (was Bitbucket Cloud, causing auth errors for GitHub-only users)

v1.18.0

March 11, 2026

Added

PR Search & Filter — Click the search icon (🔍) in the Pull Requests panel to filter by title, author, PR number, or branch name. Active filter is shown in the panel header. Clear with one click (✕)

v1.17.0

March 11, 2026

Added

Copilot CLI Agent (Experimental) — Review PRs using your GitHub Copilot subscription. Supports Claude, GPT, and Gemini models through Copilot — no separate API key needed. Install with npm i -g @github/copilot, then enable in Settings → Models. Note: Copilot CLI is v1.0.x and its API may change between updates

v1.16.2

March 10, 2026

Updated

Changed

AI Models — Updated default models: Gemini 3.1 Pro, GPT-5.4, removed legacy models. Added Gemini 3 Flash, Gemini 3.1 Flash Lite, GPT-5.4 Pro, GPT-5.3 Instant, o4-mini

Fixed

Profile switching — Instant visual feedback when switching profiles (optimistic UI update)

v1.16.0

March 10, 2026

Added

.gitautoreview.md Config File — Drop a .gitautoreview.md file in your repo root to give the AI project-specific context, rules, and conventions. No settings needed — the file content is automatically included in every review
Published Comments Collapse — Published comments now collapse into a togglable section. Auto-collapsed when more than 3 comments to keep the review panel clean
Model Selector in Sidebar — Switch AI models directly from the review panel without opening Settings. Multi-select supported

v1.15.2

March 10, 2026

Fixed

Workspace folder matching — Fixed reviews failing when your local folder name doesn't exactly match the repository name (e.g., "casino-frontend-monorepo" for "casino-frontend"). Now uses fuzzy matching and git remote URL fallback
Loading overlay contrast — Agent activity log text is now brighter and more readable against the dark overlay

v1.15.0

March 9, 2026

Added

Code Quality Assessment — Every review now scores your code across 6 categories (1-5 scale): Design, Clarity, Error Handling, Complexity, Testability, and Production Readiness. Includes actionable recommendations and strengths. Toggle categories individually in Settings → Review tab
Acceptance Criteria Verification — AI checks whether your PR meets acceptance criteria from three sources: Jira ticket, PR description, and manual entry. Each criterion is marked as addressed or not addressed
Review Profiles — Save different configurations (AI model, rules, provider settings) as named profiles and switch between them instantly
Status bar indicator — See your active profile at a glance. Click to switch profiles via Quick Pick
Workspace auto-binding — Bind profiles to workspace folders so the right configuration loads automatically when you open a project
Profile management — Create, duplicate, rename, and delete profiles from the Settings panel (General tab)
Quality Assessment Settings UI — Enable/disable quality scoring and individual categories from the Settings Review tab with a custom quality prompt option

Fixed

CLI timeout handling — Agent timeout (10 min) now shows a clear error instead of silently falling back to API. No retry, no wasted time
Acceptance criteria accuracy — Plain bullet lists in PR descriptions are no longer mistakenly treated as acceptance criteria. Only explicit ## Acceptance Criteria sections with checkboxes trigger verification
Activity log layout — Agent activity log now fills the entire sidebar panel width and height
Telemetry event type — Fixed error reporting event type for backend compatibility

v1.14.0

March 8, 2026

Added

Agent tools — The Claude CLI agent now checks lint results, test coverage, existing PR comments, and related test files during reviews for deeper analysis
Real-time progress — Agent reports what it's doing as it works: running lint, checking coverage, reading teammate comments

Fixed

Duplicate comments — Fixed an issue where review comments appeared twice when selecting a PR from the sidebar
Agent log filtering — MCP progress messages no longer leak into the Output channel as warnings

v1.13.0

March 7, 2026

Added

Claude CLI Agent — Review PRs using your Claude Code subscription instead of an API key. The agent reads your project files for deeper, context-aware reviews
Automatic fallback — If Claude CLI isn't installed or authenticated, reviews seamlessly fall back to the Claude API
Real-time agent activity — Watch the agent work in real time: reading files, searching patterns, analyzing code. Full activity log with dark overlay
CLI settings — Configure custom CLI command, see installation status and version in the Settings panel

Improved

Changed

Settings header — Refreshed design with animated owl logo

v1.12.5

March 5, 2026

Improved

Changed

Reduced token usage — Optimized prompts use fewer tokens per review, lowering costs across all AI providers
Claude 4.6 models — Default Claude model updated to Sonnet 4.6 for better review quality. Claude Opus 4.6 and Sonnet 4.6 added to model selector
Smaller extension — Extension package reduced from 6 MB to under 500 KB by removing unnecessary assets

Fixed

AI response parsing — Fixed rare issue where code suggestions containing markdown code blocks could cause the AI response to be parsed incorrectly

v1.12.2

March 5, 2026

Fixed

GitLab Self-Managed — Fixed diff loading failures on some self-managed GitLab instances by automatically falling back to an alternative API when the primary endpoint is unavailable

v1.11.0

March 1, 2026

Improved

Changed

Extension Stability — Improved internal error handling and recovery across the extension to reduce unexpected issues during reviews and settings usage

v1.10.0

February 27, 2026

Improved

Changed

Settings Panel Stability — Settings panel is now significantly more reliable. Errors are caught and shown with a clear banner offering "Copy Logs" and "Reload" options instead of a blank screen
Faster Settings Load — Optimized how the settings panel loads and renders, reducing startup time

Fixed

Settings Panel Blank Screen — Fixed an issue where the settings panel could show a completely blank screen on certain configurations
Empty Red Boxes — Removed phantom red validation boxes that appeared in settings UI when no errors were present
Special Characters in Settings — Fixed incorrect display of & and " characters in repository names and other user-provided fields

v1.9.0

February 25, 2026

Added

GitLab Settings UI — Added GitLab PAT configuration section to the Integrations tab. Configure your Personal Access Token and instance URL (cloud or self-managed) directly from the settings panel with instant validation
CSP-Safe Inline Handlers — Replaced 9 CSP-blocked inline onchange handlers with delegated event listeners, fixing broken selects and checkboxes in the settings webview
Load from Git Button Fix — Fixed "Load from Git" button to correctly fetch git config instead of re-sending current settings

Fixed

Jira Acceptance Criteria Flow — Complete overhaul of Jira AC integration: recursive ADF text extraction, proper customfield_ filtering, hardBreak support, AC field missing warning, OpenAI/Gemini AC support, URL normalization, AC verification preserved in multi-model merges and chunked reviews, innerHTML→textContent XSS fix, maxComments clamping

v1.8.6

February 24, 2026

Fixed

Repository Limit Display — Fixed "0 / NULL" display for team/business tier customers where unlimited repository limits (NULL in database) were not handled correctly. The Add Repository button was incorrectly disabled for unlimited plans
Division by Zero Guard — Added fallback for edge case where repository limit could be zero, preventing potential display issues
GitLab Option in Add Repository — Added GitLab as an option in the Add Repository and Team Repository platform menus

v1.8.3

February 23, 2026

Fixed

Team Tab Visibility Race Condition — Fixed a race condition where the Team tab wouldn't appear for Whop customers because license info was fetched before identity sync completed, reading tier as "free". Team tab now shows from the initial settings load regardless of licenseUpdated timing. Added error resilience and a post-sync retry as safety net

v1.8.2

February 23, 2026

Fixed

Team Tab After Purchase — Settings panel now shows the Team tab immediately after a Whop purchase completes, without requiring a VS Code restart. Previously, the tier change was detected but the Settings webview wasn't notified

v1.8.1

February 23, 2026

Added

Team Jira Sharing — Admin configures Jira once, all team members inherit the settings automatically. Credentials stay local per user for security
Smart Provider Defaults — Switching review modes now only enables AI providers that have a configured API key, preventing "key not configured" errors

Improved

Changed

Team Member Experience — Team members now see correct tier, usage limits, and feature access. Previously, members could appear as free tier locally despite being on a team plan
Review Mode Persistence — Customized AI provider selections (e.g., OpenAI-only) are now correctly preserved when switching between Fast, Thorough, and Custom modes

v1.8.0

February 22, 2026

Added

GitLab Integration — Full support for GitLab Cloud and self-hosted instances. Merge request listing, diff fetching, comments, file content, and status mapping — all through the unified platform interface
GitLab Self-Hosted SSL — Custom CA certificates and skip-SSL-verification options for corporate GitLab instances behind proxies
Platform Auto-Detection — Extension now automatically detects whether a repository is GitHub, GitLab, or Bitbucket from the git remote URL. Supports SSH, HTTPS, OAuth token URLs, and enterprise/self-hosted instances
Retry Customization — shouldRetry predicate now supports three-way control: false (block retry), true (force retry on any error), undefined (use default behavior). Useful for platform-specific retry policies

Improved

Changed

AI Response Handling — Improved normalization of AI model responses. Category and severity fields returned as arrays are now correctly extracted. Invalid enum values default gracefully instead of breaking the review
Gemini Model Update — Updated default Gemini model to gemini-2.5-flash across the extension
Diff Processing Architecture — Extracted diff splitting, file filtering, chunking, and validation into a standalone utility module for better maintainability and testability
Network Resilience — Retry logic now correctly handles custom predicates without conflicting with the built-in retryable error check
Jira Setup Instructions — Settings panel now shows complete list of required API token scopes for Jira integration, including granular scopes for scoped tokens

Fixed

GitLab Merged MR Diffs — Fixed diff fetching for merged/closed merge requests where GitLab returns null diff references
TypeScript Build — Fixed spread type error in settings panel that could cause compilation issues in strict mode

v1.7.0

February 21, 2026

Improved

Changed

Team Collaboration — Shared repositories and shared API keys are now fully operational end-to-end. Team admins set keys and repos once, members get them seamlessly on any machine
Settings Webview — Replaced inline event handlers with delegated data-action pattern for better Content Security Policy compliance
CodeLens Lifecycle — Active issue decorations are properly cleared on extension deactivation, preventing stale indicators in the editor

v1.6.0

February 20, 2026

Added

Shared Repositories — Team admins can add shared repository configurations from the Settings Team tab. All team members automatically see shared repos in the PR sidebar alongside their personal repos, with deduplication and (Team) label
Shared API Keys — Team admins can set shared AI provider keys (Anthropic, Gemini, OpenAI) for the entire team. Members without personal keys automatically fall back to the team key — no more "set your own token" prompts
Team Repos & Keys Management UI — New "Shared Repositories" and "Shared API Keys" sections in the Settings Team tab with add/remove controls for admins

Fixed

Model Display Name — Extension now shows correct model names (e.g., "GPT-5" instead of "GPT-4") by looking up the model version registry instead of hardcoding
Cold Cache on First Activation — Team repos and shared API keys now work on first extension activation. Previously, empty cache on startup silently skipped team features until next reload

Security

Hardened Team API — Improved authorization checks and input validation across team management endpoints

v1.5.0

February 20, 2026

Added

PR Preview Panel — Click a PR in the sidebar to see details (title, author, branches, description, changed files with diff stats) without triggering an AI review. No more accidental token usage
Start Review Button — Explicit "Start Review" button in the preview panel. AI review only happens when you ask for it
Smart Cached Reviews — If you've already reviewed a PR, clicking it again shows cached results instantly — no preview step, no waiting
Stale Review Detection — When a PR changes after your review, a warning banner appears with a one-click "Re-review" option
Review Cancellation — Cancel an in-progress review via the progress notification. AI API calls are aborted mid-stream, no partial results leak through
Post-Cancel Recovery — After cancelling, you return to the PR preview (or cached results for re-reviews) — never a blank screen
Cancellation Cooldown — 2.5-second cooldown after cancel prevents accidental double-reviews
Continue Review — If you dismiss the cancel confirmation, a new progress notification appears so you never lose track of a running review

Fixed

Deprecated Model Fallback — Extension now auto-detects deprecated AI models in saved presets and falls back to defaults. No more 404 errors from old model IDs like claude-3-5-haiku
Telemetry Event Naming — Fixed review_cache_check event using underscore instead of dot convention. Added 5 missing event types to backend allowlist (review.cache_check, review.cancelled, email.gate_shown, email.registered, email.gate_dismissed)
Data Sanitizer Allowlist — Added prId, hasCachedReview, cachedAt, stage, durationMs properties to telemetry sanitizer so cache/cancel analytics actually reach the backend
Marketplace URL — Fixed VS Code Marketplace feedback link pointing to wrong publisher ID

v1.4.0

February 18, 2026

Added

Team Management — Create teams, invite members, and manage seats from the Settings Team tab. Owners and admins can invite by email, cancel pending invites, and remove members
Team Analytics — Team-level review stats with total reviews, hours saved, active members count, trend indicators (up/down/flat), and per-member leaderboard
Invite System — Invite codes with copy-to-clipboard for manual sharing. Auto-detect pending invites on extension startup with accept/dismiss notification
Auto Team Creation — Purchasing a Team or Business plan via Whop automatically creates a team and adds you as owner

Fixed

Gauge SVG — Acceptance rate semicircle gauge now renders correctly for all percentages. Previously, values above 50% caused the arc to draw outside the visible area
Backend Auth — Improved server-side authentication flow to match the extension's auth model
Schema Alignment — Backend functions updated to match production database schema

v1.3.1

February 18, 2026

Fixed

Whop Plan Limits — Users who purchased plans through Whop were incorrectly shown "limit reached" because the extension only checked the legacy license key table. Now falls back to the identity-synced tier from the users table, so Whop-purchased plans (Starter, Developer, Team, Business) are correctly recognized
Register Email Button — "Register Email" button in History and Insights tabs now triggers the email gate and refreshes the UI after registration completes

v1.3.0

February 18, 2026

Added

Email Identity — Register your email to unlock enhanced features. Quick, non-intrusive gate after your 3rd anonymous review with clear value exchange
Review History — Browse past reviews in the Settings panel History tab. See date, repo, issue count, and categories for every review. Paginated with infinite scroll
Personalized Insights — New Insights tab shows your top issue categories, severity distribution, and trends over time. Track your code quality improvement
Tiered Free Plan — Anonymous users get 3 reviews/day; register email for 10/day. Upgrade for more
Whop Checkout Integration — Upgrade buttons now open Whop checkout with your email pre-filled for seamless purchasing
Post-Purchase Sync — After checkout, the extension polls for tier changes every 30 seconds and updates your plan within minutes
Grace Period Notifications — If your subscription lapses, you get a 7-day grace period with a clear notification and one-click resubscribe

Improved

Changed

5 Pricing Tiers — Expanded from 3 to 5 paid tiers (Starter, Developer, Team, Business) to better match different needs
License Cache — Reduced from 24-hour to 1-hour TTL for faster tier change detection
GDPR Consent — Separate privacy consent for email storage, independent from telemetry consent

Security

Whop Webhook Verification — Cryptographic signature validation with replay attack protection
Secure Email Storage — Email stored in OS keychain via VS Code SecretStorage, never in plaintext

v1.2.4

February 16, 2026

Fixed

Event Tracking — Fixed 5 feedback/suggestion event types that were rejected by the backend due to naming mismatch (dot vs underscore convention). Events suggestion_approved, suggestion_rejected, review_completed, review_feedback, and suggestion_feedback now arrive correctly in the database
Event Buffer — Malformed events (400 Bad Request) are now dropped instead of retried forever, preventing stale events from blocking the entire queue

v1.2.1

February 15, 2026

Improved

Changed

Native Icons — Replaced emoji characters with VS Code codicon icons across all UI surfaces (settings panel, review webview, edit comment panel, CodeLens, QuickPick). Icons now render consistently across all OS and themes
Consistent Severity Indicators — Severity icons in CodeLens and comment threads now use standard VS Code theme icons ($(error), $(warning), $(info), $(pass)) instead of colored emoji circles

v1.2.0

February 14, 2026

Added

Review Feedback — Rate each review with thumbs up/down directly in the review panel. Helps us understand what's valuable
Suggestion Feedback — Mark individual suggestions as accurate or false positive with inline thumb icons on each issue card
Your Review Stats — Personal analytics dashboard in Settings showing acceptance rate gauge, category breakdown, and time period filtering
Smart Milestone Prompts — Non-intrusive feedback prompt after your 10th review, with "Don't ask again" option that syncs across machines
Privacy-First Telemetry — All analytics respect VS Code telemetry settings. Transparent event manifest in telemetry.json

Improved

Changed

Feedback Icons — Replaced text labels with clear thumb up/down SVG icons for better UX
Publish Button — Fixed text contrast on green background for better readability

Security & Privacy

Security

Data Sanitization — All telemetry events are sanitized: no file paths, no code content, no PII
Offline Resilience — Events queue locally when offline and sync when connection returns
Dual Consent — Telemetry requires both VS Code global setting and extension-specific opt-in

v1.1.0

January 18, 2025

Added

🎉 GitHub Integration — Full support for GitHub repositories! Review PRs directly from GitHub with the same powerful AI analysis
Smart Issue Deduplication — AI no longer reports the same issue multiple times; similar findings are automatically merged into one actionable comment
Published Comment Awareness — Regenerating reviews skips issues you've already addressed, keeping your workflow clean

Improved

Changed

Precise Comment Placement — Comments now appear exactly on the problematic line, not before or after
Cleaner UI — Issue badges wrap gracefully when there are many labels; "Go to code" button stays visible

Fixed

Line Number Accuracy — Resolved edge cases where AI comments pointed to wrong lines in chained method calls
Badge Overflow — Labels no longer push buttons off-screen on narrow panels

v1.0.7

January 17, 2025

Added

AI Confidence Scoring — Each review issue now shows a confidence percentage, helping you prioritize what matters most
Smart PR Filtering — Focus on your work: filter PRs by author with one click
Git Blame Context — AI understands code history to provide more relevant feedback
Bearer Token Auth — New authentication option for Bitbucket Repository Access Tokens
Repository Limits — Visual counter shows how many repos you can add on your plan
Credential Validation — Helpful warnings when Bitbucket credentials are misconfigured

Improved

Changed

Better Error Messages — Clear, actionable guidance when API calls fail (with "Open Settings" button)
Faster Reviews — Performance optimizations and code quality improvements
Sidebar Layout — Long file paths now wrap correctly instead of overflowing

Fixed

Pagination State — Fixed edge case where "Load More" button could show incorrect state
Settings Sync — Repository counter updates immediately when license changes
Type Safety — Resolved potential undefined value issues throughout the codebase

v1.0.1

January 4, 2025

Added

Edit Panel — Rich editing experience with live preview for AI-generated comments
Jira Settings — Configure Jira integration directly from the settings panel
Usage Dashboard — Track your review usage with visual progress indicators
Team Management — Invite team members and manage roles (Pro tier)