Changelog

Track every feature, improvement, and fix as we build the future of Glue-powered development intelligence.

v0.11.0

Latest

2026-02-09

Added

  • •
    Ticket Chat Agent
    Conversational AI assistant for deep ticket exploration
  • •
    Natural language Q&A about ticket requirements, specifications, dev plans, and codebase context
  • •
    Full ticket context passed in system prompt (ticket details, triage analysis, feature specification, dev plan, parent ticket info)
  • •
    Multi-turn conversation support with history preservation and follow-up questions
  • •
    Session management
    save and load previous chat sessions for each ticket
  • •
    Real-time SSE streaming for responsive AI interactions
  • •
    10 specialized tools
    semantic_search, search_symbols, get_method_body, get_type_definition, get_config_constants, find_api_endpoints, get_symbol_call_graph, search_features, get_schema, get_ticket_details
  • •
    Thread root tracking for conversation continuity (parentSessionId + threadRootId)
  • •
    Floating chat panel with session history dropdown
  • •
    User-friendly progress indicators with tool message mapping
  • •
    Built on Claude Agent SDK with Sonnet 4.5
  • •
    New API routes
    `/api/tickets/[ticketId]/chat/stream` (POST), `/api/tickets/[ticketId]/chat/conversations` (GET), `/api/tickets/[ticketId]/chat/sessions/[sessionId]` (GET)
  • •
    Reusable chat components
    TicketChatBadge, TicketChatDropdown, TicketChatList, TicketChatItem
  • •
    User-Friendly Tool Message Mapping
    Technical tool names automatically mapped to friendly progress messages
  • •
    Created `/lib/toolMessageMapper.ts` with 30+ tool name mappings
  • •
    Maps technical names to user-friendly messages
  • •
    `semantic_search` → "Searching codebase"
  • •
    `get_method_body` → "Reading code"
  • •
    `find_api_endpoints` → "Finding API routes"
  • •
    `Bash` → "Checking under the hood"
  • •
    `Skill` → "Bringing in the specialist"
  • •
    And 25+ more mappings for all Stella MCP tools and built-in tools
  • •
    Applied to ALL progress indicators
    triage, dev plan, chat, notification banners
  • •
    Includes mcp__ prefixed versions for seamless compatibility
  • •
    Graceful fallback to original tool name if unmapped

Changed

  • •
    Ticket Detail UI Improvements
  • •
    Goals and Features now display above Description in OverviewTab (better information hierarchy)
  • •
    Added loading state checks to all notification banners (prevents premature error displays)
  • •
    Chat panel now 30% width when active with floating scroll-to-bottom button
  • •
    Progress indicators no longer show tool counts and iteration counts
  • •
    Completion messages simplified
    "Triage completed" (removed technical stats)
  • •
    Ticket Chat Threading Model
  • •
    Separated `parentSessionId` (immediate previous session) from `threadRootId` (first session in conversation)
  • •
    `parentSessionId` changes with each message in the thread
  • •
    `threadRootId` stays constant for entire conversation thread
  • •
    Proper threading support for follow-up questions and multi-turn conversations
  • •
    Agent SDK Integration
  • •
    Added `ticket_chat` tool set to AGENT_TOOL_SETS in stella-mcp-server.ts
  • •
    System prompt includes comprehensive ticket context (no tools needed for ticket data)
  • •
    "New Chat" properly clears both parentSessionId and threadRootId
  • •
    Session metadata includes threadRootId for conversation continuity

Fixed

  • •
    Banner Display Timing
    All inline notification banners now wait for data to load (`!isLoading` check added)
  • •
    Scroll Behavior
    Instant scroll to bottom when loading existing chat sessions (no animation)
  • •
    Session Threading
    Fixed duplicate variable declarations causing incorrect session IDs in payload
  • •
    Message Order
    Chat messages now display chronologically (oldest to newest, newest at bottom)

v0.10.0

2026-02-09

Added

  • •
    Agent SDK Tool Expansion
    Complete coverage of all code symbol types across all agents
  • •
    Created `get_type_definition` tool for retrieving interfaces, type aliases, and enums
  • •
    Created `get_config_constants` tool for retrieving configuration variables and constants
  • •
    Expanded `get_method_body` to include classes, UI components, and hooks (was functions/methods only)
  • •
    Added both new tools to all 7 agents
    Q&A, Feature Enrichment, Feature Discovery, Gap Analysis, Comprehensive Gap, Ticket Triage, Dev Plan
  • •
    Updated agent prompts with clear tool usage guidance (when to use each tool)
  • •
    Updated all skills (ticket-triage, spec-generation, ask) with new tool instructions
  • •
    Added `get_schema` tool to 5 additional agents (Q&A, Feature Enrichment, Feature Discovery, Gap Analysis, Comprehensive Gap)
  • •
    Added schema fallback instructions
    "if empty, infer schema from code"
  • •
    Language-agnostic descriptions supporting TypeScript, Python, Java, Go, and all indexed languages
  • •
    Bulk Operation Modal Improvements
    Independent ticket loading
  • •
    Batch Triage Modal now loads up to 1000 IMPORTED tickets when opened (independent of page filters)
  • •
    Bulk Approval Modal now loads up to 1000 PM_REVIEW tickets when opened
  • •
    Bulk Dev Plan Modal now loads up to 1000 PM_APPROVED tickets when opened
  • •
    All modals show loading spinner during fetch
  • •
    Removed dependency on parent page's ticket list, pagination, and filters
  • •
    Ticket Relationship Features
  • •
    New API
    `/api/projects/[id]/tickets/by-goal-or-feature` - fetch tickets sharing same goal/feature
  • •
    RelatedTicketsTooltip component for hover displays of related tickets
  • •
    Ticket detail APIs now include goal_name field in responses
  • •
    Roadmaps Page Enhancements
  • •
    PLAN_READY notification banner with MCP implementation guide
  • •
    Filter options now loaded from backend API (stages, priorities, assignees, external statuses with counts)
  • •
    Goals sorted by ticket count (largest goals first)
  • •
    Replaced icons with hex logo for batch operation CTAs
  • •
    View mode support (list/goals/kanban) with conditional pagination
  • •
    Ticket Detail Improvements
  • •
    New AdditionalInfoTab replacing ActivityTab for better information architecture
  • •
    ImplementationTab enhanced with dev plan step tracking
  • •
    OverviewTab redesigned with improved layout and goal/feature display

Changed

  • •
    Agent Tool Coverage
    Expanded from 2 symbol types (function, method) to 10 types
  • •
    `get_method_body`
    function, method, class, component, hook
  • •
    `get_type_definition`
    interface, type, enum
  • •
    `get_config_constants`
    variable, constant
  • •
    Excluded
    namespace, module (rarely useful)
  • •
    Tool Parameters
    Removed redundant `symbol_type` parameter from new tools (tool name defines what it fetches)
  • •
    Agent SDK Session Tracking
    Added SessionSerializer to dev-plan-agent-sdk and ticket-triage-agent-sdk for full session persistence
  • •
    Tickets API
    Conditional pagination based on view mode (list view only), filter options now included in all responses
  • •
    TicketDetailSheet
    Streamlined from 274+ lines, removed redundant code, improved tab structure

v0.9.0

2026-02-09

Added

  • •
    Pipeline UX Overhaul
    Complete redesign of the project creation pipeline experience
  • •
    Live progress timer showing elapsed time per phase
  • •
    Dynamic value links — completed steps now link directly to their results
  • •
    Live progress counters showing items indexed/discovered in real-time
  • •
    Auto-chaining — pipeline phases automatically trigger the next step on completion
  • •
    Proper hover inversion on pipeline cards (links, buttons, icons all invert)
  • •
    User-facing step labels instead of internal names
  • •
    Polling loop with redirect to project page on completion
  • •
    Empty States Redesign
    Storytelling-driven empty states across the entire app
  • •
    Dashboard, workspaces, workspace detail pages all have prominent amber CTAs
  • •
    Gap Analysis page now shows a "Run Gap Analysis" empty state with 4-step feature grid
  • •
    Your Insights page with welcome message and guided next steps
  • •
    Outcome-driven messaging that sells the value before asking for action
  • •
    Trial plan with 1M tokens auto-created on signup
  • •
    New Brand Identity
    Hex graph logo across the entire application
  • •
    Replaced all text-based logos with the new hex logo
  • •
    Updated favicon and app icons
  • •
    Consistent branding in header, sidebar, and footer
  • •
    Security Headers
    Added enterprise-grade security headers
  • •
    X-Frame-Options (DENY), Strict-Transport-Security (HSTS with preload)
  • •
    X-Content-Type-Options, Referrer-Policy, Permissions-Policy
  • •
    Footer & Navigation
    Expanded site-wide navigation
  • •
    Footer now renders on all marketing pages with links to About, Benefits, Blog, Authors, Changelog
  • •
    Header navigation expanded with About, Blog, Changelog links
  • •
    Blog post sidebar always visible with tags and "All Posts" link
  • •
    Blog Post Metadata
    Full OpenGraph, Twitter Card, and canonical URL metadata on every blog post
  • •
    Content Management
    New utility scripts for blog/tag coverage checking and content fixes

Changed

  • •
    Design System
    Full dark theme standardization across 8+ pages using documentation pattern
  • •
    All cards, list items, and interactive elements have proper color inversion on hover
  • •
    Admin pages audited for dark theme compliance
  • •
    Hover text visibility and export button styling fixed
  • •
    Enrichment concurrency
    Increased from 3 to 5 parallel feature enrichments
  • •
    Amplitude tracking
    Comprehensive tracking added for every page route

Fixed

  • •
    Pipeline auto-chain
    Gap analysis now waits until all features are enriched before starting
  • •
    Enrichment pipeline
    Properly filters unenriched features, bounded concurrency, retry logic
  • •
    Feature discovery/enrichment
    Consolidated into a single service layer
  • •
    Primary language detection
    Now picks the first development language, not just the most bytes
  • •
    FK violation on signup
    Fixed foreign key constraint error during new user registration
  • •
    ProjectFilterDropdown
    Proper 401 handling when session expires
  • •
    Trial plan SQL
    Uses IF NOT EXISTS to prevent duplicate plan creation errors
  • •
    Sidebar background
    Now matches page background using oklch(0.145 0 0)
  • •
    Blog soft 404s
    Posts with empty content now return proper HTTP 404 instead of 200

v0.8.0

2026-02-07

Added

  • •
    Complete Blog & Content Management System
    Full CMS infrastructure with 50 published blog posts
  • •
    New DB tables
    `content_authors`, `content_tags`, `content_post_tags`, `content_pages` (migration 016)
  • •
    Blog listing page at `/blog` with category sidebar, tags, keyword cloud
  • •
    Blog post page at `/blog/[slug]` with sticky related posts sidebar, author links, tag links
  • •
    Author pages at `/authors` (listing) and `/author/[id]` (profile with posts)
  • •
    Tag pages at `/tag/[tag]` with keyword fallback for backward compatibility
  • •
    Static pages at `/about`, `/benefits` rendered from `content_pages` table
  • •
    Landing page at `/ai-work-breakdown-structure-generator` (ChatGPT referral traffic)
  • •
    All blog queries JOIN `content_authors` for author profiles
  • •
    50 Published Blog Posts
    Content covering 4 strategic pillars
  • •
    Pillar 1
    "The Understanding Tax" (flagship essay + supporting posts)
  • •
    Pillar 2
    "Tribal Knowledge Is a Liability" (knowledge management)
  • •
    Pillar 3
    "AI Tools Need Context" (Copilot/Cursor/Claude Code gaps)
  • •
    Pillar 4
    Technical Deep Dives (Louvain clustering, MCP, embeddings vs knowledge graphs)
  • •
    16 tags with 120 tag-post associations
  • •
    3 author profiles (IDs 56, 76, 120 matching legacy indexed URLs)
  • •
    Every post has 2 internal links, "pre-code intelligence" mention, and Glue product reference
  • •
    Audience split
    23 engineer, 21 CTO, 3 executive, 3 PM
  • •
    SEO & Crawl Infrastructure
  • •
    Dynamic `sitemap.xml` querying all content tables (posts, authors, tags, pages, ebooks)
  • •
    `robots.txt` with all AI crawler user-agents (GPTBot, ClaudeBot, PerplexityBot, etc.)
  • •
    Full meta tags on every page
    title, description, keywords, openGraph (with siteName), twitter cards
  • •
    `<link rel="canonical">` headers for all pages
  • •
    Schema.org `TechArticle` structured data on blog posts
  • •
    3 Static Content Pages
    About, Benefits, AI Work Breakdown Structure Generator
  • •
    Content stored in `content_pages` table with markdown rendering
  • •
    Full SEO metadata from database

Changed

  • •
    Middleware
    Broadened matcher to catch all routes for www redirect, auth checks only on app routes
  • •
    next.config.ts
    Added redirects for nested slug (`/blog/migration-comparison/lovable-dev-comparison`), typo (`/picth-vc`), and canonical headers
  • •
    Blog listing
    Added tags section in sidebar, keywords link to `/tag/` routes, author names from JOIN
  • •
    API routes
    `/api/blog` and `/api/blog/[slug]` now JOIN `content_authors`, return tags
  • •
    generate-content.ts
    Creates/links `content_authors` records automatically
  • •
    setup-content.ts
    Runs both migration 011 and 016
  • •
    Migration
    `016_content_authors_tags_pages.sql` (5 new tables/columns)
  • •
    Seed files
    10 SQL files for content, tags, links, and fixes
  • •
    14 new route files across 7 marketing page directories
  • •
    10 modified source files (blog pages, API routes, middleware, robots, sitemap, scripts)
  • •
    All 50 posts have internal cross-links per Hitesh's SEO advice (2 links per post)
  • •
    6 thin posts expanded to 2000+ characters
  • •
    Build
    Compiles successfully

Fixed

  • •
    Soft 404 → Proper 404
    Blog `[slug]` page now calls `notFound()` instead of returning 200 with "Post not found"
  • •
    www → non-www redirect
    Middleware-level 301 redirect for all `www.glue.tools` requests
  • •
    Blog post UI/UX
    Related posts moved to sticky right sidebar, badges made borderless, transition durations added, code block borders fixed to `border-[0.5px]`

v0.7.0

2026-02-07

Added

  • •
    Theme System
    Dark/light mode with amber accent color throughout the app
  • •
    Design System Compliance
    Dashboard, sidebar, badges, hover states all theme-aware

v0.6.1

2026-02-07

Added

  • •
    Batch Dev Plan Generation
    New modal for bulk ticket dev plan generation
  • •
    Select up to 20 PM_APPROVED tickets for batch dev plan generation
  • •
    Search tickets by title, description, or ID
  • •
    Filter tickets by roadmap with group view option
  • •
    Real-time progress tracking with SSE (Server-Sent Events)
  • •
    Individual ticket status indicators (pending, in-progress, completed, failed)
  • •
    Progress bar showing X/Y tickets completed
  • •
    Parallel processing
    3 tickets processed simultaneously for 3x faster throughput
  • •
    All selection controls disabled during processing
  • •
    Dev Plan Agent SDK
    Reusable dev plan agent service
  • •
    New service
    `dev-plan-agent-sdk.ts` following established patterns
  • •
    Generates implementation steps with file-level tasks based on actual architecture
  • •
    Full Claude Agent SDK integration with session tracking
  • •
    Cost and token analysis per dev plan generation
  • •
    SSE streaming support for real-time progress updates
  • •
    Dev Plan Streaming APIs
    New endpoints for single and bulk dev plan generation
  • •
    `/api/projects/[projectId]/tickets/[ticketId]/dev-plan/stream` - Single ticket dev plan with SSE
  • •
    `/api/projects/[projectId]/tickets/bulk-dev-plan` - Batch dev plan generation (POST)
  • •
    Parallel execution
    3 tickets at a time for optimal performance
  • •
    Extended timeout
    20 minutes for long-running operations
  • •
    SSE events
    batch_started, ticket_started, ticket_completed, ticket_failed, batch_complete, done
  • •
    Inline Dev Plan Progress
    Real-time dev plan generation notifications
  • •
    Live progress indicators show when dev plan is being generated
  • •
    Color-coded status badges (yellow=generating, green=success, red=error)
  • •
    Current tool being used displayed in real-time
  • •
    Toast notifications on success/failure
  • •
    Import Stream Improvements
    SSE-based progress tracking for integrations
  • •
    New streaming endpoints for ClickUp, Jira, and Linear imports
  • •
    Real-time progress updates during bulk import operations
  • •
    Hooks
    `useClickUpImportStream`, `useJiraImportStream`, `useLinearImportStream`
  • •
    Floating action bar with live progress visualization
  • •
    Event-driven architecture replacing polling-based updates
  • •
    Roadmaps Empty State
    Dedicated empty state component for tickets page
  • •
    Shows when no tickets exist in the project
  • •
    Actionable CTA buttons for importing from ClickUp/Jira/Linear
  • •
    Visual guidance with icons and clear messaging

Changed

  • •
    URL Structure Update
    Tickets page renamed to Roadmaps
  • •
    URL changed from `/projects/[projectId]/tickets` to `/projects/[projectId]/roadmaps`
  • •
    Page label shows "TICKETS" while title shows "Roadmaps"
  • •
    Breadcrumbs updated
    "Dashboard > Projects > Roadmaps"
  • •
    Sidebar navigation updated with new route structure
  • •
    Old `/tickets` URL redirects to `/roadmaps` for backwards compatibility
  • •
    Batch Modal UX Simplification
    Removed completion view from batch modals
  • •
    Both BatchTriageModal and BulkDevPlanModal now auto-close after completion
  • •
    Modal closes automatically after 1 second and page refreshes to show updated tickets
  • •
    Cleaner UX flow without intermediate completion screens
  • •
    Users see updated ticket statuses immediately on page reload
  • •
    Modal Titles
    Standardized naming across batch modals
  • •
    "Batch Dev Plan Generation" → "Batch Dev Plan" for consistency
  • •
    Matches "Batch Triage" naming convention
  • •
    New component
    `BulkDevPlanModal.tsx` with SSE progress tracking
  • •
    New service
    `dev-plan-agent-sdk.ts` (~450 LOC)
  • •
    New API routes
    `bulk-dev-plan/route.ts`, `dev-plan/stream/route.ts`
  • •
    New hooks
    `useClickUpImportStream.ts`, `useJiraImportStream.ts`, `useLinearImportStream.ts`
  • •
    New component
    `FloatingActionBarWithProgress.tsx` for import operations
  • •
    Modified files
    AppSidebar.tsx, BatchTriageModal.tsx, TicketRow.tsx, TicketCard.tsx, ImplementationTab.tsx
  • •
    Parallel processing
    `PARALLEL_LIMIT = 3` for batch operations
  • •
    SSE format
    `event: name\ndata: {...}\n\n`
  • •
    Dev plan API uses MCP tools
    semantic_search, search_symbols, get_method_body, find_api_endpoints, get_symbol_call_graph
  • •
    Build
    ✅ Compiles successfully

v0.6.0

2026-02-06

Added

  • •
    New Project Creation Wizard (`/projects/new`)
    Full-page 4-step wizard built from scratch
  • •
    Step 1 - Select Repositories
    Two-column picker with search, organization filter, add/remove
  • •
    Step 2 - Configure Branches
    Branch dropdown per repo with auto-selection of main/master
  • •
    Step 3 - Name Project
    Project name with validation, debounced duplicate checking, optional description, summary card
  • •
    Step 4 - Review & Launch
    6 pipeline cards showing what AI agents will do, grouped by execution phase
  • •
    Phase 1
    Code Indexing (~10 min)
  • •
    Phase 2
    Commit History + Schema Detection (parallel, ~5 min)
  • •
    Phase 3
    Feature Discovery (~10 min)
  • •
    Phase 4
    Feature Enrichment (~30 min)
  • •
    Phase 5
    Gap Analysis (~10 min)
  • •
    Create Project button triggers project + workspace creation and kicks off the full pipeline
  • •
    Progress states
    preview, creating, created, error with retry support
  • •
    Sidebar Integration
    "+" button in Projects section header (both collapsed and expanded modes) navigates to `/projects/new`

Changed

  • •
    Project Detail Page
    Removed `project_code_map_status` database query and onboarding redirect logic - now renders project dashboard directly
  • •
    Dashboard Empty State
    Replaced scrolling marketing journey with simple "No project selected" message
  • •
    NoWorkspacesState
    Simplified from wizard-launching component to simple redirect link
  • •
    New files
    7 (CreateProjectWizard directory + route page)
  • •
    Deleted files
    29 (onboarding, wizard, modal, hooks, API routes)
  • •
    Modified files
    ~15 (consumer cleanup)
  • •
    Dead code preserved
    `createProjectWithMultipleRepositories` in projectService.ts (no callers, will be reused by new API)
  • •
    TypeScript
    Zero compilation errors
  • •
    `project_code_map_status` table audited across both codebases (Node.js agents don't update it - fix pending)
  • •
    Pipeline split
    Stella (Java) handles indexing/commits/schema, Node.js Agent SDK handles feature discovery/enrichment/gap analysis
  • •
    Node.js agents
    `feature-discovery-agent-sdk.ts`, `feature-enrichment-agent-sdk.ts`, `comprehensive-gap-agent.ts`
  • •
    Status tracking
    `project_code_map_status` table is the single source of truth - each service updates its own column
  • •
    No polling needed
    frontend reads the table once on page load, services write as they complete

v0.5.0

2026-02-05

Added

  • •
    Product Demo Screenshots
    New optimized screenshots for product showcase
  • •
    Documentation page full view screenshot
  • •
    Feature catalog listing view
  • •
    Gap Atlas comprehensive view
  • •
    Product demo GIF and MP4 video files
  • •
    Test frame for video generation workflow
  • •
    Static Feature Catalog Demo
    Tour page rebuilt as interactive demo
  • •
    10 mock features with realistic names and descriptions
  • •
    Live search filtering by feature name and description
  • •
    Statistics dashboard showing total features, enriched count, files mapped, and API routes
  • •
    Hover effects with light background transitions
  • •
    Enriched feature indicators with green checkmarks
  • •
    File count and route count tooltips
  • •
    Designed for demo/marketing purposes on tour route

Changed

  • •
    Tour page (`/projects/[id]/documentation/[featureId]/tour/[workflowIndex]`) repurposed as Feature Catalog demo
  • •
    Sticky header on documentation pages for better UX
  • •
    Layout optimization for documentation tour route
  • •
    Modified files
    ComplexityTab.tsx, RefactorCandidatesTab.tsx, ReleasesTab.tsx, FileHealthTab.tsx
  • •
    Tour page completely rewritten with static mock data (250 lines)
  • •
    Added file extension variant logic to tour page file loader
  • •
    New product screenshots in `/public/screenshots/optimized/`

Fixed

  • •
    Code Health UI Fixes
    Corrected invalid Tailwind CSS opacity syntax across multiple components
  • •
    ComplexityTab
    Fixed nested opacity values (e.g., `bg-white/50/20` → `bg-white/10`)
  • •
    RefactorCandidatesTab
    Fixed hover state opacity syntax
  • •
    ReleasesTab
    Fixed multiple instances of invalid opacity nesting in buttons and badges
  • •
    FileHealthTab
    Fixed styling inconsistencies
  • •
    All components now use valid single-level opacity values
  • •
    Tour Page File Loading
    Improved fuzzy file matching for code references
  • •
    Automatically tries alternative file extensions (.ts ↔ .tsx, .js ↔ .jsx)
  • •
    Handles files with or without 'src/' prefix
  • •
    Reduces "file not found" errors when tour steps reference slightly incorrect paths

v0.4.9

2026-02-05

Added

  • •
    Custom Tooltip Component
    Polished tooltip UI for goals and features
  • •
    Title and description displayed on separate lines
  • •
    Badge showing relevance percentage and confidence
  • •
    Black background with white border matching design system
  • •
    Portal-based rendering to appear above all UI elements (z-index
    999999)
  • •
    Shows "This ticket contributes to project goal" for goals
  • •
    Shows "This ticket is related to product feature" for features
  • •
    Ticket List Enhancements
  • •
    Internal status badge moved to top right beside ticket title
  • •
    Roadmap name displayed with yellow folder icon beside ticket ID
  • •
    Dot separator (·) between ticket ID and roadmap name
  • •
    Similar tickets now open in new tabs with proper URL parameters
  • •
    Batch Triage Modal
    New modal for bulk ticket triage operations
  • •
    Select up to 20 tickets from IMPORTED stage
  • •
    Search tickets by title, description, or ID
  • •
    Filter tickets by roadmap
  • •
    Group view by roadmap with collapsible sections
  • •
    Real-time progress tracking with SSE (Server-Sent Events)
  • •
    Individual ticket status indicators (pending, in-progress, completed, failed)
  • •
    Progress bar showing X/Y tickets completed
  • •
    Parallel processing
    3 tickets triaged simultaneously
  • •
    All selection controls disabled during processing
  • •
    Batch Triage API
    New endpoint for bulk ticket processing
  • •
    `/api/projects/[projectId]/tickets/batch-triage` (POST)
  • •
    Processes up to 20 tickets per batch
  • •
    Parallel execution (3 tickets at a time) for faster processing
  • •
    SSE streaming with events
    batch_started, ticket_started, ticket_completed, ticket_failed, batch_complete, done
  • •
    20-minute timeout for long-running operations
  • •
    Sequential error handling with per-ticket retry
  • •
    Reusable Triage Agent SDK
    Refactored triage logic into reusable service
  • •
    New service
    `ticket-triage-agent-sdk.ts` following established patterns
  • •
    Two exported functions
    `runTicketTriageAgentStream()` for single tickets, `runTicketTriageAgentSDK()` for batch operations
  • •
    Preserves exact same prompts and triage workflow
  • •
    Markdown extraction and cleaning utilities
  • •
    SSE writer helper class for consistent event formatting

Changed

  • •
    Badge Styling Improvements
  • •
    Internal status badges show only colored icon + text (no background/border)
  • •
    External status badges have `bg-white/10` background
  • •
    Triage details tab badges restored with backgrounds for better contrast
  • •
    Badge component now supports `plain` prop for borderless variant
  • •
    Ticket Details Panel UI Polish
  • •
    Hover effects on all triage details sections (white background with dark text)
  • •
    Improved contrast for badges, text, and similar ticket cards on hover
  • •
    Similar ticket cards have lighter greyish background (`bg-white/[0.08]`)
  • •
    Darker hover state for individual ticket cards
  • •
    All section elements respond to group hover state with smooth transitions
  • •
    Similar Tickets Section
    Only individual ticket cards turn white on hover (not entire section)
  • •
    Roadmaps Filter
    Removed roadmaps button beside list/board view switcher
  • •
    Single Ticket Triage Route
    Simplified to use reusable triage agent SDK (reduced from ~435 lines to ~80 lines)

Fixed

  • •
    Contrast Issues
    All text elements in triage details tab now have proper contrast on white hover backgrounds
  • •
    Similar Tickets Navigation
    Fixed URL parameter from `ticketId` to `ticket` to match page expectations
  • •
    Tooltip Positioning
    Tooltips now render in document body portal to prevent z-index issues with sidebar
  • •
    New component
    `Tooltip.tsx` with React portal and dynamic positioning
  • •
    New component
    `BatchTriageModal.tsx` with SSE progress tracking
  • •
    New service
    `ticket-triage-agent-sdk.ts` with streaming and non-streaming variants
  • •
    New API route
    `batch-triage/route.ts` with parallel processing
  • •
    Modified files
    TicketRow.tsx, TicketStatusBadge.tsx, TriageDetailsTab.tsx, ActivityTab.tsx, tickets/page.tsx
  • •
    Color palette
    `#FBBF24` (yellow), `#ededed` (hover white), `#050505` (dark)
  • •
    Transition
    `transition-all duration-300` for smooth animations
  • •
    Batch processing
    `PARALLEL_LIMIT = 3` for optimal throughput
  • •
    SSE format
    `event: name\ndata: {...}\n\n`

v0.4.8

2026-02-05

Added

  • •
    Home9 - Linear-Inspired Homepage
    Complete redesign with cinematic product storytelling
  • •
    Hero section
    "CODE TO CONTEXT" with "Your code knows everything" messaging
  • •
    6 feature showcases with real product screenshots mapped to actual URLs
  • •
    Interactive Code Tour, Feature Catalog, Ticket Lifecycle, Work Items, AI Chat, Gap Atlas
  • •
    Scroll-triggered parallax effects with perspective transforms
  • •
    "The Difference" before/after comparison (6 weeks → 4 days)
  • •
    "5-Minute Onboarding" section with 3-step flow visualization
  • •
    Deep feature descriptions with bullet points and live demo links
  • •
    Color-coded badges (blue, purple, red, orange, yellow, green) per feature
  • •
    Stats panels showing real metrics (16 steps, 28 iterations, 21 gaps, etc.)
  • •
    Rounded corners, subtle borders, hover states throughout
  • •
    8 product screenshots copied to `/public/screenshots/`
  • •
    Feature Documentation Page Redesign
    Storytelling-focused layout
  • •
    Sticky sidebar navigation with workflows and section links
  • •
    Interactive file preview with syntax highlighting
  • •
    Comprehensive code navigation and data exploration
  • •
    System prompt storage and workflows display
  • •
    Snake_case field support for relationships API

Changed

  • •
    APIRoutesSection
    Support snake_case fields from relationships API
  • •
    Sidebar Navigation
    Use relationship stats for accurate section counts
  • •
    File Preview
    Added syntax highlighting for better code readability

Fixed

  • •
    Relationships Junction Table
    Use correct data structure for files and routes
  • •
    Sidebar Scroll Navigation
    Improved console logging for debugging
  • •
    Files modified
    Home9.tsx, feature documentation page components, APIRoutesSection
  • •
    Added 8 new product screenshots with proper naming convention
  • •
    Framer Motion scroll hooks
    `useScroll`, `useTransform` for parallax
  • •
    Perspective transforms
    `rotateX`, `rotateY`, `scale` animations
  • •
    Easing curves
    `[0.16, 1, 0.3, 1]` for premium feel
  • •
    Build
    ✅ Compiles successfully

v0.4.7

2026-02-05

Added

  • •
    Triage Details Tab
    New dedicated tab showing structured triage analysis
  • •
    Classification, severity, priority, clarity badges with color coding
  • •
    Confidence percentage display
  • •
    Frequency pattern and impact assessment sections
  • •
    Similar tickets list with relationship types (EXACT_DUPLICATE, HIGHLY_RELATED, SIMILAR_AREA)
  • •
    Auto-opens when triage details exist
  • •
    Clickable similar tickets to navigate between related issues
  • •
    Next Step Banner for Spec Approval
    PM_REVIEW stage now shows approval banner
  • •
    Green banner with "Next Step
    Approve Specification" messaging
  • •
    Approve button moved from actions bar to banner (matches triage UX pattern)
  • •
    Removed "Request Changes" and "Reject" buttons
  • •
    Side Panel Animations
    Chat messages now animate on appearance
  • •
    Framer-motion fade-in with slide-down effect
  • •
    Staggered animation for multiple messages
  • •
    Messages display in reverse order (newest on top)

Changed

  • •
    Ticket Triage Configuration
    Improved spec generation reliability
  • •
    Increased `maxTurns` from 20 to 60 to allow agent to complete and output final spec
  • •
    Explicitly set model to `claude-sonnet-4-5-20250929` (Sonnet 4.5)
  • •
    Fixed tool count tracking (was always showing 0 tools called)
  • •
    Panel State Management
    Better handling when switching between tickets
  • •
    Resets all state (ticket data, active tab, artifacts, messages) when clicking similar tickets
  • •
    Prevents stale data from previous ticket appearing in new ticket view

Fixed

  • •
    Empty Final Output Issue
    Triage was completing without generating specification text
  • •
    Root cause
    Agent was running out of turns (20) before producing final output
  • •
    Solution
    Increased maxTurns to 60, allowing full exploration + spec generation
  • •
    Files modified
    TicketDetailSheet.tsx, TriageDetailsTab.tsx, TicketActions.tsx, triage/stream/route.ts
  • •
    Added framer-motion animations for improved UX
  • •
    Removed Link component from similar tickets (now uses onClick handler)
  • •
    Build
    ✅ Compiles successfully

v0.4.6

2026-02-04

Added

  • •
    Node.js Onboarding Orchestration
    Moved Feature Discovery, Enrichment, and Gap Analysis from Java to Node.js
  • •
    New `OnboardingOrchestrator` class coordinates the full onboarding pipeline
  • •
    `OnboardingStatusService` manages `project_code_map_status` table updates
  • •
    `OnboardingSSEService` broadcasts real-time events via PostgreSQL NOTIFY/LISTEN
  • •
    New SSE endpoint
    `/api/projects/[id]/onboarding-stream` for frontend updates

Changed

  • •
    Java Backend Scope Reduced
    Java now only handles Code Indexing and Commit Fetching
  • •
    Feature Discovery → Node.js `runFeatureDiscoveryAgentSDK()`
  • •
    Feature Enrichment → Node.js `runFeatureEnrichmentAgentSDK()` (batches of 3)
  • •
    Gap Analysis → Node.js `runComprehensiveGapAgent()`
  • •
    Fail-Fast Pipeline
    If any step fails, subsequent steps don't run
  • •
    Frontend SSE
    `useProjectStream` now uses Node.js endpoint instead of Java proxy
  • •
    New files
    `src/lib/onboarding/` (types, StatusService, SSEService, Orchestrator)
  • •
    New endpoint
    `src/app/api/projects/[projectId]/onboarding-stream/route.ts`
  • •
    Modified
    `src/app/api/projects/onboard/route.ts` - now orchestrates Node.js agents
  • •
    SSE events include `agentType` and `statusType` for frontend compatibility
  • •
    Checkpoint/resume support for enrichment batch failures
  • •
    Build
    ✅ Compiles successfully with no type errors

v0.4.5

2026-02-04

Added

  • •
    Centralized Tool Architecture
    Complete refactoring of agent tools
  • •
    Each tool now has ONE file in `tools/definitions/` (single source of truth)
  • •
    Tool files export
    `TOOL_NAME`, `TOOL_DESCRIPTION`, `TOOL_USAGE`, `createTool()`
  • •
    `getAgentToolConfig(agentType)` provides standardized configuration for all agents
  • •
    `generateToolPromptSection()` creates consistent tool documentation for system prompts
  • •
    Predefined tool sets
    `qna`, `feature_enrichment`, `feature_discovery`, `gap_analysis`, `comprehensive_gap`

Changed

  • •
    Standardized Agent Tool Configuration
    All 5 SDK agents now use `getAgentToolConfig()`
  • •
    Uniform tool parameter documentation in system prompts
  • •
    Consistent MCP server creation across agents
  • •
    Shared tool descriptions and usage examples
  • •
    Deleted 16 Duplicate/Legacy Files
  • •
    10 legacy tool class files (search-symbols.ts, get-method-body.ts, etc.)
  • •
    2 legacy agent files (qna-agent.ts, gap-analysis-agent.ts)
  • •
    1 unused types file (tools/types.ts)
  • •
    3 documentation files from agents folder
  • •
    Removed `AbstractTool` class from base.ts
  • •
    Tool definitions reduced from scattered 600+ lines to focused single-responsibility files
  • •
    Agent imports simplified
    `import { getAgentToolConfig } from '../tools/stella-mcp-server'`
  • •
    New directory structure
    `tools/definitions/` contains 7 tool files + index.ts
  • •
    Build
    ✅ Compiles successfully with no type errors

v0.4.4

2026-02-04

Added

  • •
    Changelog Modal Timeline
    Homepage modal now shows curated highlights for the most recent releases (v0.4.0+), giving teams a quick history without leaving the app.

Changed

  • •
    Enhanced Logging
    Added detailed debug logs for conversation threading
  • •
    SessionSerializer logs thread root resolution, history length, and first question
  • •
    QnA Agent logs full prompt being sent to Claude SDK (for debugging)
  • •
    Better visibility into parent/thread relationship resolution
  • •
    Unified Agent Tools
    All Claude SDK agents now share a single Stella MCP tool server so every tool (semantic_search, find_api_endpoints, etc.) has the same schema, description, and behavior across Q&A, enrichment, discovery, and gap-analysis flows.
  • •
    Chat page now tracks both `currentSessionId` (from URL) and `latestSessionId` (in-memory state)
  • •
    Thread root ID flows
    URL → frontend → API → Agent SDK → SessionSerializer
  • •
    Parent session ID flows
    `latestSessionId` state → API → creates proper linked list
  • •
    Build
    ✅ All TypeScript type errors resolved

Fixed

  • •
    Conversation Threading
    Complete overhaul of session linkage system
  • •
    Added `latestSessionId` state tracking in chat UI to maintain proper parent chain
  • •
    Fixed `parent_session_id` linkage - now points to immediate previous message instead of thread root
  • •
    Added `threadRootId` parameter passing from URL to backend (previously missing)
  • •
    QnA Agent now receives correct `threadRootId` for history loading
  • •
    Session API returns latest session ID in thread for proper follow-up threading
  • •
    SessionSerializer improved with database query for thread root when not provided
  • •
    Duplicate Answer Display Bug
    Simplified message loading logic
  • •
    Removed complex `content_blocks` extraction from iterations
  • •
    Chat messages now use simple `question` and `answer` fields directly from `agent_sessions`
  • •
    Eliminated failed deduplication logic that was causing double display
  • •
    Cleaner separation
    `messages` array for chat display, `iterations` array for debug panel
  • •
    Conversation History Loading
    Fixed history not being available to agent
  • •
    `loadThreadHistory()` now correctly resolves `threadRootId` from config or database
  • •
    Added comprehensive logging for thread history debugging
  • •
    History prompt properly includes all previous Q&A pairs in chronological order
  • •
    Agent receives full conversation context for follow-up questions
  • •
    Thinking Status UI Overflow
    Fixed thinking text pushing content below input area
  • •
    Added proper text wrapping with `break-words` to prevent horizontal overflow
  • •
    Improved icon alignment with `flex-shrink-0` for multi-line thinking text
  • •
    Removed text truncation - all thinking content displays fully
  • •
    Added bottom margin to separate from subsequent messages

v0.4.3

2026-02-04

Added

  • •
    Interactive File References in Documentation
    Click any file reference to view code
  • •
    Smart path resolution (handles partial paths, full paths, with/without `src/`)
  • •
    Line number and line range highlighting (e.g., `file.ts
    96-779`)
  • •
    Fuzzy file search fallback for partial filenames
  • •
    Syntax-highlighted code viewer modal with line numbers
  • •
    Auto-detection of class names (e.g., `FileIndexingService` → tries `.ts`, `.tsx`, `.java`, `.py`)
  • •
    Enhanced Workflow Flow Pages
    Completely redesigned `/flow/[flowIndex]` pages
  • •
    Right sidebar with persistent context (trigger, execution flow, tech stack)
  • •
    Vertical execution flow visualization with numbered steps and arrows
  • •
    Smart content parsing
    extracts numbered lists, SQL queries, API endpoints
  • •
    Markdown rendering with GlueMarkdown component for consistent styling
  • •
    Tech context bar showing framework patterns, database operations, key functions
  • •
    Collapsible SQL query blocks with syntax highlighting
  • •
    Smart architecture overview parser (detects and structures numbered lists)
  • •
    Documentation Page Stats Enhancement
  • •
    Total features with breakdown (discovered, suggested, manual)
  • •
    Enriched features count vs pending enrichment
  • •
    Sublabel support in StatsCard component
  • •
    Export functionality (Markdown, HTML, JSON)
  • •
    Workflow Detail Component Library
    New reusable component system
  • •
    `WorkflowDetail/FileReference.tsx` - Interactive file viewer (150 lines)
  • •
    `WorkflowDetail/ContextSidebar.tsx` - Metadata sidebar (180 lines)
  • •
    `WorkflowDetail/WorkflowStepCard.tsx` - Individual step cards (120 lines)
  • •
    `WorkflowDetail/StepContentRenderer.tsx` - Smart content parsing (95 lines)
  • •
    `WorkflowDetail/parseStepContent.ts` - Parsing utilities (125 lines)

Changed

  • •
    Feature Enrichment Agent Prompt
    Fixed JSON output format
  • •
    Now generates correct `{ format
    "workflows_v2", content: {...} }` structure
  • •
    Steps use `name` and single `description` field (not separate `action`, `system_behavior` fields)
  • •
    Added optional `trigger`, `flow_chain`, and `layer` fields
  • •
    Workflows now match UI expectations without adapter code
  • •
    Workflow Flow Page
    Refactored from 968 lines to 384 lines
  • •
    Extracted reusable components
  • •
    Improved code organization and maintainability
  • •
    Better separation of concerns
  • •
    StatsCard Component
    Added sublabel support for secondary information
  • •
    9 orphaned files deleted
  • •
    `page-old.tsx` - deprecated file browser
  • •
    3 `.backup` API route files
  • •
    3 legacy Dashboard components (TaskList, SprintProgress, QuickActions)
  • •
    `RepositoriesSheet.tsx` - unused component
  • •
    `FEATURE_DISCOVERY_ENHANCED.json` - test data
  • •
    5 unused functions from `workspaceService.ts` (~175 lines)
  • •
    `getWorkspacesByProject()`
  • •
    `getWorkspaceByProjectAndBranch()`
  • •
    `updateWorkspace()`
  • •
    `deleteWorkspace()`
  • •
    `getWorkspaceFileStatsByProject()`

Fixed

  • •
    TypeScript strict null checks in feature-enrichment-agent-sdk
  • •
    TypeScript implicit any parameter types in database query maps
  • •
    Missing null checks in feature-discovery-agent-sdk
  • •
    File reference extraction to support line ranges (`
    96-779`)
  • •
    ContextSidebar conditional rendering (hides when no metadata)
  • •
    Interactive file viewer uses `/api/workspaces/[workspaceId]/files/content` endpoint
  • •
    Smart parser handles
    numbered lists, SQL queries, API endpoints, file references, markdown headers
  • •
    Component architecture enables reuse across documentation views
  • •
    Build
    ✅ Compiles successfully

v0.4.2

2026-02-04

Added

  • •
    Feature Enrichment Agent (TypeScript SDK)
    Complete port from Java backend
  • •
    Generates comprehensive user journey workflows and technical documentation
  • •
    Uses Claude Agent SDK with MCP tools (semantic_search, search_symbols, get_method_body, get_symbol_call_graph, search_features)
  • •
    Session tracking with full cost and token analysis
  • •
    SSE streaming support for real-time progress
  • •
    Saves to `feature_catalog.user_journeys` as JSONB
  • •
    Comprehensive logging
    context loading, iterations, tool calls, results, costs
  • •
    Batch processing support (3-5 features in parallel)

Changed

  • •
    All Feature Enrichment Endpoints Now Use TypeScript
  • •
    `POST /api/features/enrich-journeys` - Single feature enrichment (was Java proxy)
  • •
    `POST /api/workspaces/[workspaceId]/features/enrich` - Batch enrichment (was Java proxy)
  • •
    `POST /api/workspaces/[workspaceId]/generate-documentation` - Documentation generation with enrichment (both sync and async modes)
  • •
    Removed Java backend dependencies for enrichment
  • •
    10-20 minute timeout per feature (complex workflows)
  • •
    Real-time progress notifications during batch processing
  • •
    Gap Analysis API Enhanced
    Added dashboard integration fields
  • •
    Returns
    competitors[], criticalCount, quickWinsCount, topCategory, topGap
  • •
    Fixed impact score thresholds (80/60 instead of 8/5)
  • •
    Dashboard card now displays complete competitive intelligence

Fixed

  • •
    Feature discovery route syntax errors (missing if/else structure)
  • •
    Duplicate exports in agents/index.ts
  • •
    Missing pool import in features/discover route
  • •
    Missing enrichJourneys variable declaration
  • •
    TypeScript strict null checks in feature-discovery-agent-sdk
  • •
    Feature Enrichment Agent
    ~670 LOC with comprehensive logging
  • •
    Follows same pattern as Q&A and Gap Analysis agents
  • •
    Full integration in 3 API endpoints
  • •
    Build
    ✅ Compiles successfully

v0.4.1

2026-02-04

Changed

  • •
    Model Update
    All agents now use claude-sonnet-4-5-20250929 (Sonnet 4.5) instead of claude-sonnet-4-20250514
  • •
    Updated default model in qna-agent.ts, qna-agent-sdk.ts, gap-analysis-agent.ts
  • •
    Updated session-serializer.ts default
  • •
    Updated migration 013 defaults
  • •
    Updated scripts/generate-content.ts
  • •
    Kept old model in pricing table for backwards compatibility

Fixed

  • •
    TypeScript Build Errors
    Fixed type checking issues in comprehensive-gap-agent.ts
  • •
    Corrected SDK message type handling (message.type doesn't include 'tool_use' directly)
  • •
    Fixed assistant message content block iteration pattern
  • •
    Removed duplicate result handling code
  • •
    Fixed property access
    phase_1_product_identity.competitors instead of phase_1_competitors

v0.4.0

2026-02-03

Added

  • •
    Claude Agent SDK Integration
    Complete agent framework built on official Claude Agent SDK
  • •
    Full TypeScript implementation in `src/lib/agent-sdk/`
  • •
    MCP server integration for custom code intelligence tools
  • •
    Session serialization with database persistence
  • •
    Token tracking with prompt caching support (90% cost reduction)
  • •
    Cost calculation per session (input/output/cache tokens with USD pricing)
  • •
    Q&A Agent (SDK Version)
    Natural language codebase exploration agent
  • •
    Multi-turn conversations with history preservation
  • •
    6 specialized tools
    semantic_search, search_symbols, get_method_body, find_api_endpoints, get_symbol_call_graph, search_features
  • •
    SSE streaming for real-time progress updates
  • •
    Database persistence
    agent_sessions, agent_iterations, tool_executions tables
  • •
    Conversation threading via thread_root_id
  • •
    Error tracking and recovery
  • •
    API endpoints
    `/api/workspaces/ask` and `/api/workspaces/ask/stream`
  • •
    Comprehensive Gap Analysis Super-Agent
    Automated competitive intelligence system
  • •
    4-Phase Pipeline running in single agent (~2-3 minutes total)
  • •
    Phase 1
    Product Identity & Competitor Detection (15s) - Auto-detects product type and top 3 competitors via WebSearch
  • •
    Phase 2
    Competitor Feature Discovery (30s) - Discovers top 20 features per competitor from changelogs and product pages
  • •
    Phase 3
    Gap Analysis (90s) - Analyzes YOUR codebase for each competitor feature, calculates coverage %
  • •
    Phase 4
    Feature Correlation & Deduplication (5s) - Merges duplicate features across competitors
  • •
    Fire-and-forget execution (non-blocking background job)
  • •
    Evidence-based rationale with source URLs from WebSearch
  • •
    Impact scoring algorithm (0-100) based on competitive saturation + market demand
  • •
    Stores validated gaps in `competitor_comparison_features` table
  • •
    API endpoint
    `POST /api/projects/[projectId]/gap-analysis/analyze`
  • •
    UI integration
    "ANALYSE MARKET GAPS" button on gap-analysis page
  • •
    10 Code Intelligence Tools
    Ported from Java backend with identical SQL queries
  • •
    `semantic_search` - AI-powered conceptual search using OpenAI embeddings
  • •
    `search_symbols` - Find functions, classes, methods by name pattern
  • •
    `get_method_body` - Retrieve actual source code implementations
  • •
    `find_api_endpoints` - Discover REST API routes
  • •
    `get_symbol_call_graph` - Trace callers and callees
  • •
    `search_features` - Search feature_catalog
  • •
    `get_schema` - Database schema extraction
  • •
    `get_type_relationships` - Type hierarchies
  • •
    `get_file_structure` - List files with symbols
  • •
    `get_ticket_details` - Fetch ticket/task information
  • •
    Skills Framework
    User-facing shortcuts for common agent tasks
  • •
    `/ask` - General Q&A about codebase
  • •
    `/explain` - Detailed feature explanation with user flow + technical implementation
  • •
    `/find` - Locate code by concept or keyword
  • •
    `/trace` - Function call graph analysis
  • •
    Skills defined as filesystem artifacts in `.claude/skills/` directory
  • •
    Each skill has SKILL.md with description, tools, and prompt template
  • •
    Session Serializer
    Complete agent session tracking and persistence
  • •
    Tracks sessions, iterations, tool calls with millisecond timing
  • •
    Token usage tracking (input, output, cache_read, cache_creation, thinking tokens)
  • •
    Cost calculation using model-specific pricing (Sonnet 4, Opus 4, Haiku)
  • •
    Conversation history loading for follow-up questions
  • •
    Progress feed for real-time UI updates
  • •
    User quota enforcement via token balance checks
  • •
    Database Schema
    New tables for agent tracking
  • •
    `agent_sessions` - Q&A sessions with costs, token usage, success status
  • •
    `agent_iterations` - Each LLM call/response cycle with content blocks
  • •
    `tool_executions` - Individual tool invocations with timing and results
  • •
    `agent_errors` - Error tracking with severity levels
  • •
    `agent_progress_feed` - Real-time status updates for SSE streaming
  • •
    Migration
    `migrations/013_agent_sdk_serialization.sql`
  • •
    API Endpoints
    New routes for agent operations
  • •
    `POST /api/projects/[projectId]/gap-analysis/analyze` - Trigger comprehensive gap analysis
  • •
    `GET /api/projects/[projectId]/features` - Fetch features from feature_catalog
  • •
    Enhanced existing endpoints with agent integration

Changed

  • •
    Gap Analysis Page
    Now powered by Comprehensive Gap Analysis Agent
  • •
    "ANALYSE MARKET GAPS" button triggers intelligent 4-phase analysis
  • •
    Fire-and-forget execution with background processing
  • •
    Blue progress banner shows phase breakdown and estimated duration
  • •
    Auto-refresh after completion to display stored gaps
  • •
    Replaces hardcoded mock data with real agent-discovered gaps
  • •
    Homepage Redesign
    Major visual and content improvements
  • •
    New Visualizations
    Added 3 interactive journey visualizations
  • •
    TicketJourneyVisualization - Shows ticket-to-code workflow
  • •
    DocumentationVisualization - Demonstrates auto-generated docs
  • •
    GapAnalysisVisualization - Illustrates competitive gap discovery
  • •
    Before/After Section
    Simplified with large "6 wks → 3 days" comparison
  • •
    Feature Cards
    More compact design with reduced padding
  • •
    Hero Section
    Updated messaging and visual hierarchy
  • •
    Removed AudienceValue component (replaced by visualizations)
  • •
    Branding Consistency
    Updated terminology across entire application
  • •
    "AI-powered" → "Glue-powered" in all user-facing text
  • •
    Consistent messaging in email templates, SEO metadata, and UI components
  • •
    Updated homepage hero, features section, and product cards
  • •
    Schema.org structured data updated for search engines
  • •
    Chat UI Refinements
    Improved message display and spacing
  • •
    Reduced padding for more compact message bubbles
  • •
    Added `userName` prop support for personalized user messages
  • •
    Smaller metadata badges and tighter spacing
  • •
    Better JSON rendering with adaptive semantic display
  • •
    Improved responsive design for mobile
  • •
    Environment Configuration
    Added Agent SDK environment variables
  • •
    `ANTHROPIC_API_KEY` - For Claude Agent SDK (previously CLAUDE_API_KEY)
  • •
    `OPENAI_API_KEY` - For embeddings in semantic_search
  • •
    Both keys required for full agent functionality
  • •
    Tool Registry
    Centralized tool management system
  • •
    Singleton pattern for tool registration
  • •
    Tool definition generation for Claude API
  • •
    Tool execution with error handling
  • •
    Supports both legacy and SDK agent implementations
  • •
    New tables support full agent lifecycle tracking
  • •
    Composite indexes for performance (session_id, iteration_id)
  • •
    JSONB fields for flexible metadata storage
  • •
    Timestamp tracking for latency analysis
  • •
    Foreign key constraints for data integrity
  • •
    Prompt caching reduces repeat context costs by ~90%
  • •
    Tool call batching where possible
  • •
    Streaming responses for better UX
  • •
    Background job processing for long-running operations
  • •
    Connection pooling for database efficiency
  • •
    99 files modified
  • •
    38 new files created
  • •
    +11,662 insertions, -1,034 deletions
  • •
    Net
    +10,628 lines of code
  • •
    Agent SDK framework (31 files, ~3,500 LOC)
  • •
    API endpoints (2 new routes)
  • •
    UI components (3 new visualizations)
  • •
    Database migration (5 new tables)
  • •
    Documentation (4 guides + updated README/CHANGELOG)

v0.3.0

2026-02-02

Added

  • •
    Storytelling Onboarding UI
    Complete redesign of project onboarding with beautiful narrative-driven cards
  • •
    Each agent now tells a detailed story with rich prose about what it's discovering
  • •
    Real-time progress parsing from SSE events ("Enriched feature 21/25" → 84% progress)
  • •
    Completion cards show extracted metadata
    files, symbols, commits, features, duration
  • •
    Key Files section displays top files discovered (e.g., "ApiRequestLog.java, Symbol.java")
  • •
    Top Contributors section shows main developers with commit counts
  • •
    Recent Work section displays recent commit messages in full
  • •
    Discovered Features section shows feature names extracted from completion messages
  • •
    Animated progress bars with color-coded status (green=complete, blue=working)
  • •
    Grid layout with self-contained story cards (removed old split-panel view)
  • •
    Flow Detail Page Enhancements
    Added info icon modal with comprehensive feature metadata
  • •
    Architecture Overview section with full feature context
  • •
    Quick Reference panel showing
  • •
    All API routes from `routes` array
  • •
    All key files from `files` array
  • •
    All database tables from `how_it_works.tables`
  • •
    All entry points from `how_it_works.entry_points`
  • •
    Color-coded badges for different metadata types
  • •
    Accessible via info (ℹ️) icon next to workflow title
  • •
    SSL Certificate Fix
    PostgreSQL connection now properly handles GCP Cloud SQL SSL certificates
  • •
    Added `rejectUnauthorized
    false` for remote databases in development
  • •
    Fixed "unable to verify the first certificate" error during authentication

Changed

  • •
    Feature Spec Progress Calculation
    Fixed to use `checkpoint_data.enrichment.total` instead of `features_discovered`
  • •
    Now shows accurate real-time progress during feature enrichment
  • •
    Progress updates immediately from SSE events instead of waiting for polling
  • •
    Multiple fallback strategies
    SSE title parsing → checkpoint data → metadata
  • •
    Removed ALL Truncation
    No more "..." in onboarding displays
  • •
    Feature names shown in full (removed 50-char limit)
  • •
    Commit messages displayed completely (no truncation)
  • •
    All top files shown (not limited to 3)
  • •
    All contributors displayed with their commit counts
  • •
    Backend "and N more" messages properly handled
  • •
    Enhanced Metadata Extraction
    Rich data parsing from SSE completion messages
  • •
    Extracts file counts, symbol counts, commit counts, feature counts
  • •
    Parses duration strings ("Done in 3 min 58 sec")
  • •
    Identifies top files from completion summaries
  • •
    Extracts contributor names and commit counts
  • •
    Captures recent commit messages
  • •
    Parses discovered feature names from "Identified
    " patterns

Fixed

  • •
    Real-time progress tracking for feature enrichment agent
  • •
    Feature name display truncation in completion cards
  • •
    Missing metadata in agent completion summaries
  • •
    Proper handling of GCP Cloud SQL SSL connections

v0.2.0

2026-01-30

Added

  • •
    Documentation Redesign
    New two-level sidebar navigation for documentation section
  • •
    Features list with expandable workflows
  • •
    Each user workflow now has its own dedicated page at `/projects/[projectId]/documentation/[featureId]/flow/[flowIndex]`
  • •
    Main app sidebar auto-collapses when entering documentation pages
  • •
    Feature Header Enhancements
    Display category, status, source, and confidence badges in feature headers
  • •
    Architecture Overview
    Show the `user_journeys.content.overview` field on feature detail pages
  • •
    Animated @ Query Typing
    Homepage feature cards now show animated typewriter effect with rotating @ query examples

Changed

  • •
    Font Improvements
    Switched main text content from monospace to Inter font for better readability
  • •
    Long text descriptions now use regular font
  • •
    Technical values (numbers, booleans, short strings) retain monospace
  • •
    Code Health Cards
    Fixed equal height for repository health cards on code-health page

Fixed

  • •
    Various TypeScript type errors
  • •
    Fixed ref types in EmptyStateJourney component
  • •
    Fixed nullable property checks in AgentSessionView
  • •
    Fixed type error in dashboard historical click handler
  • •
    Added missing `isThread` variable in agent sessions API
  • •
    Fixed JsonRenderer export for backward compatibility
  • •
    Database Error Handling
    Agent definition API now gracefully handles missing `stella_agent_tool` table

v0.1.0

Initial Release

Glue logoGlue

Enterprise AI platform for complete codebase analysis. Built for teams of 10-500+ developers.

Product

  • About
  • Benefits
  • Dashboard

Resources

  • Blog
  • Authors
  • Changelog
  • Support

Legal

  • Privacy Policy
  • Terms of Service

Make Your Team Ready For AI Coding Tools

Claude Code•Cursor•GitHub Copilot•Windsurf•Cody

Glue provides the context layer that makes AI coding tools actually work for your team

Made with ❤️ by the Glue team • © 2026 Glue

The next revolution in software development isn't about writing code faster — it's about giving code the ability to understand itself.