Skip to content

CHANGELOG.md

Latest commit

 

History

History
565 lines (440 loc) · 25.2 KB

CHANGELOG.md

File metadata and controls

565 lines (440 loc) · 25.2 KB

Changelog

All notable changes to the vibe-log-cli project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[0.8.6] - 2025-12-05

Added

  • Custom Instructions: Personalize your AI-generated reports and standups with custom context
    • New "Custom Instructions" menu option in main menu
    • Instructions stored locally at ~/.vibe-log/instructions.md
    • Applied to daily standups and local productivity reports
    • Cloud sync support for instructions across devices
    • Add context like your role, team, projects, or preferred output style

Fixed

  • Daily standup now filters out temp projects (temp-prompt-analysis, temp-productivity-report, etc.) that were appearing in output
  • Standup tips no longer overlap - each tip now prints on a separate line instead of trying to overwrite in-place

[0.8.5] - 2025-11-22

Fixed

  • Fixed session upload metadata not being sent correctly to server

Improved

  • Improved internal tracking of upload sources

[0.8.4] - 2025-01-21

Improved

  • User-Agent header now shows actual CLI version (was hardcoded to 0.6.0)
  • CLI version now sent to server in session upload headers for version tracking

[0.8.3] - 2025-01-21

Improved

  • Session uploads no longer blocked by version updates
  • Auto-updates now happen in background after sessions are processed
  • NPX cache handling optimized to handle concurrent hook executions

Added

  • Update event logging to ~/.vibe-log/update.log

[0.8.2] - 2025-11-18

Note

  • This release reconciles the pr-8-review branch with main after the v0.8.0 release was accidentally published from a PR branch. All Cursor IDE integration features from the original PR are included in v0.8.0 and v0.8.1 below.

[0.8.1] - 2025-11-15

Fixed

  • CRITICAL: Fixed hooks overwrite bug that deleted users' custom Claude Code hooks during install/uninstall operations
    • Installing vibe-log hooks now appends to existing hooks array instead of replacing them
    • Uninstalling vibe-log hooks now filters out only vibe-log commands instead of deleting entire hook type
    • Added duplicate prevention to avoid multiple installations

Improved

  • Added comprehensive test coverage for hooks-controller (36 unit tests)
  • Removed legacy camelCase hook format support for cleaner codebase

Refactored

  • Code cleanup and improved maintainability throughout hooks system

[0.8.0] - 2025-11-07

Added

  • Cursor IDE Push-Up Challenge Integration: New hook integration for Cursor IDE's afterAgentResponse trigger
    • Automatically detects over-validation phrases in Cursor conversations
    • Supports both ASCII apostrophe (') and smart apostrophe (') characters
    • Modern Cursor format support with bubbleId-based message storage
    • Silent operation to avoid disrupting Cursor IDE workflow
  • Cursor IDE Session Sync: New cursor-upload command for uploading Cursor conversations to vibe-log
    • Date range selection options: 7 days, 30 days, or all conversations
    • Dry-run mode (--dry-run) to preview sessions before upload
    • Silent mode (--silent) for automation and scripting
    • Extracts project names from Cursor workspace databases
    • Optional file logging for troubleshooting
  • Full Cursor IDE Integration:
    • New "View Cursor IDE stats" menu option
    • Push-up challenge tracks validation phrases in both Claude Code AND Cursor IDE
    • Supports both legacy and modern Cursor conversation formats
    • Cross-platform support (macOS, Windows, Linux)
    • Time-based statistics: Shows stats for this week, last week, this month, this year
  • Added better-sqlite3 dependency for Cursor database access

Changed

  • Hook Auto-Update Mechanism: All hooks now use npx vibe-log-cli@latest instead of npx vibe-log-cli
    • Ensures users automatically get latest bug fixes and improvements
    • Applies to both Claude Code hooks and Cursor hooks

Fixed

  • Apostrophe detection for validation phrases (supports both ASCII and smart apostrophes)
  • Cursor integration timestamp initialization for accurate tracking from first enable
  • Cursor hook stdout logging silenced to prevent JSON parse errors
  • Latest message retrieval improved for modern Cursor conversation format
  • Logger file output now dynamically redirects logs after initialization

[0.7.6] - 2025-10-31

Added

  • Email Setup Integration: Push-up challenge now prompts users to enable daily emails with push-up stats and coding session summaries for standups
    • Automatically triggers cloud setup wizard if user isn't authenticated
    • Default "Yes" for streamlined onboarding
    • Only appears for non-authenticated users to avoid duplicate prompts

Fixed

  • Authentication Checks: Push-up stats sync now only attempts when user is authenticated
    • Prevents 401 Unauthorized errors during challenge enable/disable/settle/reset operations
    • Silently skips sync for local-only users
    • All sync operations check authentication status first

Improved

  • Privacy Notice Clarity: Updated privacy notice text to be more positive and clear
    • Changed from "Never store" to "Your code and personal data never leaves your machine"
    • More reassuring messaging about data privacy
  • Push-Up Menu Simplification: Streamlined push-up challenge menu display
    • Removed verbose benefit descriptions from disabled state
    • Cleaner, more focused UI
    • "The Rule" now displayed prominently in both main menu and first-time setup

[0.7.5] - 2025-10-31

Fixed

  • Push-up challenge first-time setup now correctly transitions to main menu instead of looping back to welcome screen
  • Added PUSHUP_ONLY state detection for users with only push-up challenge enabled
  • Push-up challenge status section now displays even when disabled, showing installation state
  • State refresh after first-time feature setup now works correctly

Improved

  • Added challenge and statusline installation status to push-up challenge section
  • Push-up challenge info now shown when challenge is disabled (setup instructions)
  • Cloud dashboard feature details now visible in main menu for non-authenticated users
  • Centralized menu item labels to eliminate duplication across different states
  • Increased main menu page size to prevent status sections from scrolling out of view
  • Updated statusline description text for clarity

[0.7.4] - 2025-10-30

Added

  • Push-Up Challenge Integration: New push-up challenge feature integrated into statusline
    • Tracks push-up completion during coding sessions
    • Real-time sync of push-up stats during session upload
    • Dynamic "receipt" display with better formatting and sizing

Fixed

  • Cross-Platform Compatibility: Replaced hardcoded /tmp paths with cross-platform temporary directory paths
    • Now properly uses OS-specific temp directories (Windows TEMP, Unix /tmp)
    • Eliminates path-related errors on Windows systems
  • Entry Point Configuration: Added bin/vibe-log.js entry point and updated statusline order

[0.7.3] - 2025-10-27

Changed

  • SDK Migration: Migrated from @anthropic-ai/claude-code to @anthropic-ai/claude-agent-sdk@0.1.27
    • Future-proof solution using Anthropic's actively maintained SDK
    • Better isolation with no interference from Claude Code settings
    • Maintains full API compatibility with no breaking changes

Fixed

  • "Today's standup" option now appears in LOCAL_ONLY menu for users who completed onboarding but didn't enable cloud mode
  • Standup command properly skips authentication for local-only users
  • Updated Discord invite link to correct URL

[0.7.2] - 2025-10-02

Fixed

  • Duration calculation accuracy improved using gap-based approach
  • Sessions with large time gaps between messages now accurately calculate active coding time
  • Fixed issue where total duration could exceed actual session length

[0.7.1] - 2025-10-01

Fixed

  • Test compatibility with telemetry feature
  • Chalk.hex fallback for test environments

[0.7.0] - 2025-10-01

Added

  • Daily Standup Command: New standup command to prepare for daily meetings
    • Runs locally on user's machine
    • Analyzes previous day's Claude Code sessions automatically
    • Extracts real accomplishments and features built
    • Provides business-value focused summaries
    • Includes smart suggestions for next steps
    • Anti-hallucination safeguards ensure accuracy
  • Interactive Onboarding Menu: Enhanced first-time user experience
    • Streamlined setup flow with clear guidance
  • CLI Telemetry: Anonymous usage tracking for authenticated users
    • Helps improve CLI based on actual usage patterns
    • Fully respects user privacy
  • Gamification Feature: CLI now showing streak and awarded points

Fixed

  • Session filtering now correctly excludes today's sessions from standup reports
  • Project name matching in standup duration calculations
  • Claude Code no longer creates project folders for temp directories
  • Standup command parsing with proper JSON delimiters
  • Parallel session and sub-agent dependency issues resolved
  • Time calculation consistency in standup reports

Improved

  • Temp directory cleanup for standup operations
  • Error handling and user feedback throughout the CLI
  • Configuration transmission to server for better sync

[0.6.4] - 2025-01-14

Added

  • SessionEnd Hook Support: New hook type for capturing sessions when Claude Code terminates
    • Captures sessions on clear, logout, prompt_input_exit, and other termination events
    • Full integration with existing hook management system
    • Statistics tracking for SessionEnd executions
    • Available in both global and project-specific modes
    • Testing framework support with individual and batch testing

Improved

  • Hook Management UI: Enhanced interface to include SessionEnd configuration
    • SessionEnd appears in all hook selection menus
    • Status displays show SessionEnd installation and statistics
    • Test menu includes SessionEnd hook testing option

[0.6.3] - 2025-09-12

Improved

  • Message Length Capture: Claude Code sessions now capture full message content instead of truncating at 5K characters
    • Ensures complete context preservation for analysis
    • Better analytics and insights from comprehensive session data
    • No more data loss from longer development conversations

[0.6.2] - 2025-09-10

Added

  • Git Branch Information: Sessions now capture and send the current git branch name to the API
    • Automatically extracts branch from Claude Code session directory
    • Provides better context for development activity tracking
    • Enhances analytics with branch-level insights

Improved

  • Time-Based Sync Options: Enhanced sync menu with convenient time-range options
    • Added "Last 7 days" quick sync option
    • Added "Last 14 days" quick sync option
    • Simplified workflow for regular syncing patterns
    • Better user experience for common sync scenarios

[0.6.1] - 2025-09-09

Fixed

  • Hotfix: Resolved critical issue preventing users from generating local reports in certain scenarios
    • Fixed template path resolution for NPM package distribution
    • Ensured proper access to report templates in all environments

[0.6.0] - 2025-09-05

Added

  • Template-Based Report Generation: Complete replacement of direct HTML generation with JSON→Template system
    • Claude now returns only structured JSON data
    • Fixed HTML template guarantees consistent report structure
    • All sections always appear in the correct order
    • No more missing "Prompt Quality Analysis" sections
    • Clean section titles without emojis
    • Simple stats line instead of grid cards
    • Template is the single source of truth for HTML structure
  • Version Indicators: Clear console output showing template-based system v0.6.0 is running
  • Type Safety: Full TypeScript interfaces for report data structure

Fixed

  • Report Consistency: Solved all 8 formatting issues identified in previous versions
  • Section Ordering: Executive Summary, Activity Distribution, Key Accomplishments, Prompt Quality Analysis, Project Breakdown always appear
  • Styling Issues: CSS is now predefined in template, not dynamically generated
  • Footer Problems: No unwanted footer elements in reports

Changed

  • Report Generation Flow: Claude provides data, template provides structure (separation of concerns)
  • Error Messages: More descriptive feedback during report generation process

Improved

  • Report Generation Output: Tool failures are now hidden during report generation
    • "Tool failed" messages no longer shown for expected errors (file not found, etc.)
    • Cleaner, less intimidating output during analysis
    • Error messages still visible in debug mode (VIBELOG_DEBUG=1)
    • Reduces user anxiety during the 4-5 minute generation process
  • Auto-sync menu: Simplified educational text for better clarity
    • Added shortcut command install-auto-sync for quick access
    • Added documentation link in the menu
    • Updated help text to show the new command

[0.5.3] - 2025-08-31

Fixed

  • Hook Installation: Now correctly appends vibe-log hook to existing UserPromptSubmit hooks instead of overwriting
  • Claude Sessions: Isolated all automated Claude sessions to dedicated temp directories to prevent project clutter
  • Session File Access: Fixed local report generation by copying session files to temp directory accessible by Claude
  • Project Filtering: Temp projects are now properly hidden from all project lists and menus

Added

  • Temp Directory Management: Centralized configuration for temp directories with single source of truth
  • Dark Theme: Applied consistent dark theme to local report generation for better readability

Improved

  • Project Discovery: Unified project filtering across all features using single discoverProjects function
  • Code Organization: Eliminated code duplication for temp directory handling

[0.5.2] - 2025-08-31

Improved

  • Status Line Analysis: Transformed approach from aggressive "Ship NOW!" to thoughtful questioning with specific examples
  • Session Context Awareness: Now detects first prompts, question/answer patterns, and image attachments
  • Smarter Suggestions: Provides specific test cases and edge cases to consider rather than just pushing to ship
  • Promotional Tips: Reduced frequency from 10% to 5% and disabled for first 3 messages
  • Documentation: Enhanced README to highlight status line safety features and backup/restore capability
  • User Trust: Added prominent messaging about automatic backup of existing status line configurations
  • Setup Flow: Reorganized ccusage token metrics as optional step after coach personality selection

Added

  • Session Metadata Tracking: Rich context extraction including message position and attachments
  • Thoughtful Questioning: Status line now asks clarifying questions with concrete examples
  • Safety Documentation: Clear explanation that uninstalling restores original status line configuration
  • Feature Visibility: Added backup/restore and ccusage support to feature list

[0.5.1] - 2025-08-30

Improved

  • Menu Labels: Clarified status line feature as "Prompt feedback in Claude Code"
  • UI Organization: Moved status display and learn more link to bottom of header for better flow
  • Code Quality: Centralized menu labels to follow DRY principle
  • First-time Welcome: Updated descriptions to accurately reflect status line functionality

Fixed

  • Duplicate Code: Removed redundant status display in status line menu
  • Text Accuracy: Removed "real-time" terminology as feature analyzes after prompt submission
  • Typography: Fixed "to for" typo in first-time welcome screen

[0.5.0] - 2025-08-30

Added

  • Strategic Product Advisor: Status line now acts as a proactive co-pilot that pushes users to ship faster
  • Original Mission Tracking: Remembers user's first message to maintain focus on original goal
  • Concrete Actionable Guidance: Provides specific values (timeouts, error codes) and deadlines
  • Enhanced Context Extraction: Increased conversation context from 3 to 10 turns for better analysis

Improved

  • Gordon Personality: Now sharp and business-focused with less food metaphors, creates urgency
  • Vibe-Log Personality: Supportive but pushy senior dev that helps ship with concrete deadlines
  • Prompt Analysis: More specific and action-oriented with MVP focus and "ship TODAY" mentality
  • README Documentation: Updated to reflect strategic advisor capabilities

Fixed

  • Context Extraction: Handles duplicate first messages in short conversations gracefully
  • TypeScript Compilation: Fixed potential null reference in session context extractor

[0.4.3] - 2025-08-29

Fixed

  • Status Line Analysis Timeout: Fixed critical issue where prompt analysis would timeout in Claude Code
    • Implemented background processing - hook now exits in 0.3s (was timing out at 5s)
    • Added HTML comment guard (<!--VIBE_LOG_GUARD:1-->) for clean recursion prevention
    • Removed hardcoded fallback suggestions - SDK now returns real, contextual feedback
    • Fixed SDK prompt format to ensure proper JSON response

Improved

  • Analysis Quality: Status line now provides specific, actionable suggestions based on actual prompt content
  • Loading State: Extended staleness timeout from 15s to 5 minutes for better async handling
  • Debug Logging: Enhanced SDK response logging for easier troubleshooting
  • Status Line Display: Now shows personality name (Gordon/Vibe-log/Custom) before suggestions for clarity

Technical Details

  • Background process spawns with detached: true and child.unref() for non-blocking execution
  • Analysis completes in ~6 seconds and saves to correct session file
  • Creates exactly 2 session files per prompt (user's + SDK's internal) without recursion

[0.4.2] - 2025-08-29

Changed

  • Renamed Sub-Agent: vibe-log-track-analyzer is now vibe-log-session-analyzer for better clarity
  • Breaking Change: Users need to reinstall sub-agents after this update

Fixed

  • Report Generation Performance: Optimized execution time back to 2-4 minutes
    • Removed unnecessary tools from report generator (Task, Bash, Grep, etc.)
    • Fixed batching strategy to properly limit parallel agents (MAX 9)
    • 17 sessions now correctly uses 2-3 agents instead of 17
  • Headless Mode Compatibility: Restored === REPORT START/END === markers for reliable operation
  • Sub-Agent Installer UI:
    • Shows correct count of 2 sub-agents (not 3)
    • Removed references to deprecated logs-fetcher agent
    • Updated workflow description to reflect simplified 2-phase process

Improved

  • Orchestrator Batching: Enhanced batching logic with clear examples (17 sessions = 2 agents, 25 = 3 agents)
  • Report Output Method: Reports now OUTPUT between markers instead of using Write tool
  • UI Clarity: Added "In Parallel" to Phase 1 description in sub-agent installer

[0.4.1] - 2025-08-28

Added

  • Status Line in First-Time Menu: Added "Configure prompt coach status line" option to the first-time welcome menu for improved feature discoverability during onboarding

Improved

  • Documentation: Updated README with latest features and improvements

[0.4.0] - 2025-08-28

Added

  • Real-time Status Line: Revolutionary prompt quality analyzer that provides instant feedback in Claude Code status bar
  • AI Personality System: Customizable AI personas (Gordon Ramsay, Bob Ross, etc.) for engaging prompt feedback
  • Context-Aware Analysis: Multi-turn conversation context tracking for better prompt quality assessment
  • Interactive Personality Testing: Test and preview different AI personalities before installation
  • Unified Hooks Manager: Prevent duplicate installations and conflicts between features
  • Planning Mode Detection: Automatically detect when Claude Code is in planning mode
  • Session-Specific Storage: Prevent conflicts when multiple Claude sessions are active

Improved

  • Status Line UI: Enhanced visual feedback with progress bars and actionable tips
  • Dynamic CLI Paths: Use configurable paths instead of hardcoded values
  • Prompt Analysis: Context-aware quality assessment with semantic understanding
  • Installation Flow: Streamlined status line installation with better menu integration

Fixed

  • Recursion Prevention: Robust guards to prevent infinite loops in SDK integration
  • File Lock Issues: Replaced file-based locks with prompt-based recursion guards
  • Promotional Flickering: Eliminated status line tip flickering
  • Duplicate Messages: Removed redundant installation messages
  • Creativity Preservation: Restored Claude's creative responses in personality system

[0.3.21] - 2025-08-28

Added

  • Status Line in First-Time Menu: Added "Configure Real-time prompt coach status line" option to the first-time welcome menu for improved feature discoverability during onboarding

Improved

  • User Onboarding: New users can now immediately discover and configure the AI feedback personality feature during initial setup

[0.3.20] - 2025-08-24

Added

  • Claude Model Tracking: Sessions now capture and track the specific Claude model being used (Opus, Sonnet, Haiku)
  • Comprehensive Unit Tests: Added extensive test coverage for Claude model detection and session parsing

Improved

  • Session Data: Enhanced session metadata to include model information for better analytics
  • README Documentation: Updated documentation with improved visuals and mermaid diagrams

Fixed

  • Package Distribution: Added .npmignore to optimize package size and exclude unnecessary files

[0.3.19] - 2025-08-22

Fixed

  • GitHub Actions workflows now properly configured with permissions
  • Updated artifact actions from v3 to v4 to fix deprecation warnings
  • Made lint step non-blocking in npm-publish workflow
  • Removed flaky npm verification job that failed due to propagation delays

Improved

  • Simplified npm-publish workflow for more reliable releases
  • All workflows now passing with green badges

[0.3.18] - 2025-08-22

Added

  • Package Transparency: Disabled minification for full code verifiability
  • Automated Releases: GitHub Actions workflow for npm publishing with provenance
  • Build Verification: Automated CI/CD verification across multiple platforms
  • SHA256 Checksums: Automatic generation and verification of build artifacts
  • Repository Link: Added repository field to package.json for source verification
  • Security Documentation: Comprehensive transparency section in README
  • Release Guide: Complete setup documentation for automated releases

Changed

  • Build output is now readable (non-minified) for security auditing
  • Source maps are included in npm package for debugging
  • All releases now automated via GitHub Actions for reproducibility

Security

  • Addresses community concerns about package verifiability
  • Implements npm provenance for supply chain security
  • Provides checksums for integrity verification

[0.3.17] - 2025-01-22

Improved

  • Enhanced local report analysis with activity breakdown and prompt engineering insights
  • Better async handling for package update checks
  • Cleaner repository structure

[0.3.16] - 2025-01-21

Fixed

  • Restored "Press Enter to continue" for help in first-time menu

[0.3.15] - 2025-08-20

Added

  • Sub-agent installation check before local report generation
  • Two-layer protection to ensure sub-agents are installed
  • Automatic prompt to install missing sub-agents
  • Option to continue with report generation after installation
  • Clear error messages when sub-agents are missing
  • Release process documentation in CLAUDE.md

Changed

  • Local report generation now validates sub-agent availability
  • Improved user experience with guided installation flow

[0.3.13] - 2024-08-20

Added

  • Centralized network error handling module (network-errors.ts)
  • File utilities module (file-utils.ts) for safe file operations
  • Comprehensive error type detection and categorization
  • User-friendly network error messages

Changed

  • Improved UI text clarity for cloud and local modes
  • Cloud mode now explicitly states "FREE FOREVER"
  • Local mode clarifies it uses "Claude Code with sub-agents"
  • Updated token usage estimates from 25k-100k to 10k-50k
  • Reduced time estimates from 5-15 min to 4-10 min
  • Manual sync option now clearly states "upload coding sessions"
  • Local report options specify "(using Claude sub-agents)"

Fixed

  • Eliminated ~150-200 lines of duplicate network error handling code
  • Standardized error messages across all CLI commands
  • Improved maintainability with single source of truth for errors

Refactored

  • Updated 6 files to use centralized network error utilities:
    • commands/send.ts
    • commands/auth.ts
    • utils/errors.ts
    • lib/auth/browser.ts
    • lib/api-client.ts
    • lib/ui/cloud-setup-wizard.ts

[0.3.12] - 2024-08-15

Added

  • Comprehensive language detection for Claude Code sessions
  • Session tracking metadata improvements

[0.3.11] - 2024-08-14

Changed

  • Performance improvements with increased batch sizes
  • Enhanced session tracking metadata

Fixed

  • Various performance optimizations

[0.3.10] - 2024-08-13

Added

  • Initial session tracking features
  • Core CLI functionality

For older versions, please refer to the git history.