# Changelog All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). ## [Unreleased] ## [0.9.5] - 2026-03-31 ### Fixed - **Shell Argument Escaping**: Fixed command execution to properly handle special characters in file paths - Added shell quoting for arguments containing special characters like parentheses in Next.js route groups - Prevents shell interpretation errors when running commands on files with complex names - Resolves issue #28 where hooks would fail on modern frontend project structures - **Test Command Execution**: Improved compatibility with different test runners and package managers - Fixed test command argument passing to work correctly with both npm/yarn/pnpm and direct test commands - Only uses '--' separator when running through package managers, avoiding errors with direct test executables - Ensures consistent test execution across different project configurations ### Improved - **Test File Discovery**: Enhanced test-changed hook with more robust test file detection - Added two-phase test file search: first co-located tests, then project-wide test directories - Integrated fast-glob for efficient searching in tests/, test/, and __tests__ directories - Added comprehensive JSDoc documentation for test discovery logic - Better error handling for glob operations with graceful fallbacks ## [0.9.4] - 2025-09-29 ### Fixed - **Documentation URLs**: Updated broken Claude Code documentation URLs to use current domain structure - Fixed SDK documentation URLs to include `/sdk/sdk-overview` path for accurate redirects - Updated base domain from `docs.anthropic.com` to `docs.claude.com` across documentation files - Ensures all Claude Code documentation links resolve correctly for users ### Documentation - **Agent Formatting**: Improved formatting and structure of Kafka and LoopBack expert subagents - Cleaned up verbose YAML frontmatter comments in LoopBack expert agent - Enhanced code formatting consistency and documentation structure for better readability - Streamlined agent definitions to focus on practical troubleshooting guidance ## [0.9.3] - 2025-09-28 ### Added - **Kafka Expert Subagent**: Added specialized subagent for Apache Kafka development and operations - Added `kafka-expert.md` subagent with expertise in producer/consumer configuration, topic management, cluster operations, serialization, performance tuning, and operational troubleshooting - Added symlink at `.claude/agents/kafka` for easy access - Covers Kafka errors, performance issues, consumer lag, rebalancing problems, and configuration challenges - **LoopBack Expert Subagent**: Added specialized subagent for LoopBack 4 framework development - Added `loopback-expert.md` subagent with expertise in model relationships, authentication, dependency injection, API architecture, database performance, and production deployment - Added symlink at `.claude/agents/loopback` for easy access - Handles JWT authentication errors, repository patterns, DI binding problems, and migration from LoopBack 3 ### Fixed - **File Guard Security**: Enhanced file protection service to allow absolute paths for research subagents - Fixed issue where absolute paths (like `/tmp/research_results.txt`) were incorrectly blocked by file-guard hook - Updated path protection logic in `FileProtectionService` to skip protection for absolute paths outside project root - Allows research subagents to write to temporary directories and system paths as needed - Updated tests to verify absolute paths are now allowed while maintaining protection for project files - **Link Validation**: Fixed markdown link checker configuration to handle npmjs.com URLs - Added npmjs.com URL pattern to ignore list in `.markdown-link-check.json` to prevent false positives - Added specific headers for npmjs.com requests to improve reliability when validation does occur - Resolves issues with package documentation links being flagged as broken ### Documentation - **Technical Overview**: Added comprehensive technical overview documentation with table of contents and architecture details - **Subagent Creation**: Streamlined subagent creation guide and updated model configurations - **Integration Guides**: Updated various documentation URLs to reflect current Claude Code documentation structure ## [0.9.2] - 2025-09-10 ### Fixed - **User-Scope Hook Configuration**: Fixed user-level installations to properly configure hooks in settings.json - Fixed issue where user-scope installations (`claudekit setup --user`) were not creating hook configurations in `~/.claude/settings.json` - Unified hook configuration logic by renaming `createProjectSettings` to `createSettings` and making it work for both user and project installations - Ensured both user-scope and project-scope installations now properly generate embedded hook commands (`claudekit-hooks run `) - Updated user installation workflow to create `~/.claude/settings.json` with proper hook trigger configurations - Enhanced installation consistency so hooks work correctly regardless of installation scope ## [0.9.1] - 2025-09-05 ### Changed - **Hook Extension Configuration**: Refactored hook utilities to eliminate code duplication and provide consistent file extension handling - Extracted shared extension configuration utilities into `cli/hooks/utils.ts` with `ExtensionConfigurable` interface - Added `shouldProcessFileByExtension()` function for standardized file filtering across hooks - Added `createExtensionPattern()` function for robust regex-based extension matching with proper escaping - Updated `LintChangedHook` and `TestChangedHook` to use shared extension utilities instead of duplicated logic - Enhanced type safety with guaranteed non-undefined file paths after extension validation ## [0.9.0] - 2025-09-05 ### Added - **Session-Based Hook Control**: Added comprehensive hook management system for granular control within Claude Code sessions - Added `claudekit-hooks disable [hook-name]` CLI command to disable hooks for current session - Added `claudekit-hooks enable [hook-name]` CLI command to re-enable disabled hooks for current session - Added `claudekit-hooks status [hook-name]` CLI command to show hook status with visual indicators - Added `/hook:disable`, `/hook:enable`, and `/hook:status` slash commands for use within Claude Code - Added `SessionTracker` and `SessionHookManager` classes for reliable session state management - Added fuzzy matching for hook names with partial matching and helpful suggestions - Added session state persistence in `~/.claudekit/sessions/` with atomic file operations - Added automatic transcript UUID extraction from Claude Code environment for session identification ### Changed - **Hook Architecture**: Enhanced base hook system to support session-based disable/enable functionality - Updated `BaseHook` class to automatically check session state before execution - Enhanced hook execution flow to skip disabled hooks silently without user notification - Improved session identification logic with fallback mechanisms for development scenarios - Standardized hook status reporting with emoji indicators (๐Ÿ”’ disabled, โœ… enabled) ## [0.8.14] - 2025-09-04 ### Added - **Biome Linting Support**: Added comprehensive Biome integration for modern JavaScript/TypeScript linting - Added `detectBiome()` function to detect Biome configuration files (`biome.json`, `biome.jsonc`) and dependencies (`@biomejs/biome`, `rome`) - Added `formatBiomeErrors()` utility function for consistent error formatting and actionable fix instructions - Added Biome configuration file with selective linting rules focusing on correctness and security issues - Added `lint:biome` npm script for project-wide Biome linting - Added `@biomejs/biome` as development dependency for linting capabilities ### Changed - **Tool-Agnostic Linting Architecture**: Refactored linting hooks to support multiple linting tools simultaneously - Updated `LintChangedHook` and `LintProjectHook` to run both Biome and ESLint when configured - Changed hook display names from "ESLint" to generic "Lint" to reflect multi-tool support - Updated hook dependencies from specific "eslint" to generic "linter" category - Enhanced project detection system to identify multiple linting tool configurations - Improved error reporting to show results from all configured linters with clear separation ### Fixed - **Biome Configuration**: Refined Biome setup to prevent conflicts with existing tooling - Fixed Biome formatter conflicts by disabling formatter and enabling only linter functionality - Fixed import organization conflicts by disabling Biome's import sorting to avoid ESLint/Prettier conflicts - Fixed test environment configuration with proper rule overrides for test files - Updated test expectations to reflect new hook naming and multi-tool architecture ## [0.8.13] - 2025-09-03 ### Fixed - **File-Guard Security**: Enhanced .env template file handling to allow common template patterns - Fixed issue where `.env.example`, `.env.template`, and `.env.sample` files were incorrectly blocked - Updated default sensitive patterns to include negation rules for template files (`!.env.example`, `!.env.template`, `!.env.sample`) - Improved ignore engine logic to properly handle negation patterns from custom ignore files - Enhanced file protection service to use current working directory for ignore file detection - Added comprehensive test coverage for ignore file parsing and negation pattern processing ## [0.8.12] - 2025-09-01 ### Added - **Status Command**: Added new CLI status checking functionality for tool integrations - Added `claudekit status` command with subcommand support for checking tool availability - Added `claudekit status stm` subcommand to check Simple Task Master (STM) installation and initialization status - Added status command with `--verbose` and `--quiet` options for flexible output control - Returns structured status messages: "Not installed", "Available but not initialized", or "Available and initialized" ### Fixed - **Specification Commands**: Resolved Claude Code permission errors for STM status checks - Fixed permission issues in spec decompose and execute commands when checking STM availability - Updated allowed-tools configuration to include necessary Bash commands for STM integration - Enhanced specification workflow reliability when STM tool is not available or not initialized ## [0.8.11] - 2025-08-31 ### Fixed - **Codebase Map Hook**: Resolved session handling and debugging issues - Fixed profile test session detection to prevent caching during performance testing - Enhanced manual run handling to always generate fresh output when session_id is 'unknown' - Added debug logging capabilities via `DEBUG=true` environment variable for troubleshooting - Improved session tracking logic to skip persistent flags for manual and profile test runs - Updated unit tests to reflect corrected behavior for manual execution scenarios ## [0.8.10] - 2025-08-31 ### Fixed - **File-Guard Security**: Improved git command recognition in bash command parser - Fixed false positives where legitimate git revision syntax (e.g., `@{u}..HEAD`, `origin/main..HEAD`, `HEAD~5..HEAD`) was incorrectly flagged as file access - Added `isGitRevisionSyntax()` method to properly detect and skip git revision patterns in command analysis - Enhanced pattern matching to distinguish between git refs and actual file paths - Improved security analysis accuracy while maintaining protection against actual file access attempts ## [0.8.9] - 2025-08-31 ### Changed - **Dependency Management**: Standardized file globbing library for consistency - Migrated from `glob` to `fast-glob` across all file operations for better performance - Updated `subagent-detector.ts` and `migrate-node-imports.js` to use consistent globbing - Improved build tool compatibility and faster file discovery operations - **Build Validation**: Enhanced dependency validation with source-based analysis - Completely rewrote `validate-dependencies.js` to analyze actual TypeScript imports - Improved accuracy in detecting missing and unused dependencies - Better handling of Node.js built-in modules with `node:` prefix - Validates imports directly from source files instead of built artifacts - **Test Infrastructure**: Enhanced test runner with silent mode for cleaner CI output - Added `--silent` flag to bash test runner for streamlined release preparation - Updated CI workflows to use npm scripts instead of direct bash test calls - Enhanced test framework and reporter with silent mode support - Improved release validation workflow with cleaner test output ### Fixed - **CI/CD Pipeline**: Resolved test execution issues in continuous integration - Fixed GitHub Actions workflows to properly execute tests via npm scripts - Improved test-hooks workflow reliability and consistency - Resolved bash test integration issues in automated builds - **Code Organization**: Eliminated code duplication in file-guard service classes - Extracted shared utility functions (`globToRegExp`, `untildify`) into dedicated module - Updated file-guard services to use shared utilities instead of duplicate implementations - Improved maintainability and consistency across file protection components ## [0.8.8] - 2025-08-31 ### Added - **Comprehensive Bash Security Analysis**: Enhanced file-guard hook with advanced bash command parsing - Added `BashCommandParser` class with comprehensive command analysis and path extraction - Added `SecurityHeuristicsEngine` for detecting sensitive file access patterns in pipelines - Added `FileProtectionService` with support for multiple AI ignore file formats - Added extensive bash command validation with 590+ lines of test coverage - Supports detection of sensitive file access via complex shell constructs (pipes, variables, xargs, etc.) - **Enhanced Test Infrastructure**: Expanded testing capabilities - Added comprehensive bash-specific test suite with 47 test cases - Added parallel test execution with both Vitest and bash test framework - Enhanced test scripts configuration for better CI/CD integration ### Changed - **File-guard Architecture**: Refactored file-guard hook into service-oriented architecture - Restructured monolithic file-guard hook into specialized service classes - Improved separation of concerns with dedicated parsing, security, and protection services - Enhanced maintainability and testability of security-critical code - **Bash Tool Integration**: Extended file-guard protection to cover Bash tool usage - Updated hook matcher pattern to include `Bash` tool in addition to file operations - Enhanced security coverage for AI assistant bash command execution - **Dependency Management**: Added new production dependencies for enhanced functionality - Added `fast-glob` for high-performance file pattern matching - Added `ignore` for gitignore-style pattern processing - Added `is-path-inside` for secure path boundary validation - Added `untildify` for safe home directory path expansion - Replaced `glob` with `fast-glob` for better performance ### Fixed - **Build System**: Fixed broken symlink causing CI failures - Removed broken `.claude/commands/tmp` symlink that was pointing to deleted test directory - Resolves build and deployment pipeline failures - **Documentation**: Fixed thinking-level hook configuration examples - Removed non-existent `enabled` flag from all configuration examples in documentation - Changed "superthink" to "ultrathink" to match actual implementation - Added level 0 documentation for properly disabling the hook - Updated all configuration examples to reflect actual schema and available options ## [0.8.7] - 2025-08-29 ### Fixed - **NPM Package**: Fixed test files being included in published npm package - Removed explicit `files` array from package.json that was including test directories - Added comprehensive .npmignore file with 110+ exclusion patterns - Ensures only production code (dist/, bin/, src/, README.md) is published to npm - Significantly reduces package size by excluding development and test artifacts ## [0.8.6] - 2025-08-29 ### Added - **Git Workflow Commands**: Enhanced git functionality with new commands - Added `git:checkout` command for smart branch switching with auto-completion - Added `git:ignore-init` command for initializing AI-safe .gitignore patterns - **Research Workflow System**: Multi-agent research command for parallel information gathering - Added `research` command with concurrent agent execution for comprehensive analysis - Added `research-expert` subagent with specialized research capabilities - Supports parallel research across multiple domains with structured output - **Enhanced File Security**: Expanded sensitive file protection patterns - Added comprehensive sensitive patterns module (`cli/hooks/sensitive-patterns.ts`) with 12 security categories - Added 195+ sensitive file patterns covering cloud credentials, cryptocurrencies, databases, tokens, and production data - Enhanced file-guard hook with categorized pattern management and validation utilities - Added extensive test coverage (330+ lines) for all sensitive file pattern categories ### Changed - **Command Categories**: Renamed command category system for better clarity - Changed `ai-assistant` category to `claude-setup` throughout the codebase - Updated setup wizard command grouping: "๐Ÿค– Claude Setup & Configuration" - Updated recommendation engine and component discovery to use new category names - **Release Process Integration**: Enhanced release preparation with automated quality checks - Integrated claudekit linting (`@carlrannaberg/cclint`) into release preparation workflow - Added JSON-formatted linting output parsing and error reporting - Release preparation now validates all command and agent files before publishing - **Setup Command Organization**: Improved command organization in setup wizard - Added new git workflow commands to recommended workflow group - Updated command descriptions for better clarity and consistency - Enhanced command grouping logic with better category inference ### Fixed - **Documentation Links**: Resolved broken documentation link patterns - Fixed markdown link validation patterns in configuration - Updated documentation structure and internal linking consistency - **Build Tool Recognition**: Enhanced ESBuild external dependency detection - Improved validation to properly recognize Node.js built-in modules with `node:` prefix - Fixed build tool configuration for better dependency bundling ## [0.8.5] - 2025-08-28 ### Changed - **Linting Architecture**: Refactored internal linting system to use external SDK for better maintainability - Replaced internal command linting implementation with `@carlrannaberg/cclint` SDK (v0.2.7) - Replaced internal agent linting implementation with `@carlrannaberg/cclint` SDK - Removed 720+ lines of internal linting code (`cli/lib/linters/commands.ts`, `cli/lib/linters/subagents.ts`) - Updated `lint-commands` and `lint-subagents` (now `lint-agents`) commands to use SDK-based approach - Enhanced linting with `followSymlinks: true` option for better symlink handling - Improved error counting and statistics reporting through SDK integration - **Command Naming**: Renamed `lint-subagents` command to `lint-agents` for consistency with current terminology - **Node.js Import Standards**: Fixed validation utilities to use `node:` prefix for built-in modules (child_process) ### Removed - **Internal Linting Code**: Removed internal command and agent linting implementations in favor of external SDK - Deleted `cli/lib/linters/commands.ts` (266 lines) - functionality moved to `@carlrannaberg/cclint` - Deleted `cli/lib/linters/subagents.ts` (454 lines) - functionality moved to `@carlrannaberg/cclint` - Deleted associated test files (`tests/lib/linters/subagents.test.ts`) - Reduced codebase complexity while maintaining all linting functionality ### Fixed - **Shebang Standardization**: Updated `scripts/prepare-release.sh` to use standard `#!/usr/bin/env bash` shebang ## [0.8.4] - 2025-08-27 ### Fixed - **Hook Installation**: Fixed duplicate hook installation for multi-trigger hooks - Hooks that support multiple trigger events (e.g., both `Stop` and `SubagentStop`) are now installed only once per selection - Added `hookTriggerEvents` Map to track specific trigger event assignments during setup - For individually selected hooks, only the primary (first) trigger event is used to prevent duplicates - For group-selected hooks, the hook is assigned to the specific trigger event of that group - Eliminates cluttered settings.json files with duplicate hook entries and unnecessary multiple hook executions - **Documentation Link Validation**: Fixed broken link patterns in markdown link checker configuration - Added specific replacement pattern for self-review hook documentation links - Added status code 525 to accepted alive status codes for better link validation coverage ## [0.8.3] - 2025-08-26 ### Changed - **Thinking Level Hook**: Simplified configuration and improved defaults - Simplified to 4 levels (0-3) instead of 5, as "think harder" and "ultrathink" both trigger 31,999 tokens - Changed keywords to use special Claude Code keywords: "think" (level 1), "megathink" (level 2), "ultrathink" (level 3) - Changed default thinking level from 0 to 2 ("megathink") for better reasoning out of the box - Removed unnecessary `enabled` flag - use `level: 0` to disable the hook ### Added - **Test Coverage**: Added comprehensive unit tests for thinking-level hook - 241 lines of test coverage with 493 test cases validating all hook behavior - Tests cover default behavior, level configuration, invalid input handling, and edge cases - Ensures robust validation of thinking level keyword injection ### Fixed - **Dead Code Removal**: Cleaned up unused configuration schemas - Removed `ThinkingBudgetConfigSchema` which was added but never implemented - Cleaned up references to thinking-budget throughout the codebase ## [0.8.2] - 2025-08-25 ### Added - **Hook Performance Profiling System**: Comprehensive performance analysis for hooks - Added `claudekit-hooks profile` command for measuring hook execution time and output size - Performance analysis includes execution time, character count, and estimated token usage - Support for single runs or averaged measurements across multiple iterations - Automatic detection of configured hooks from `.claude/settings.json` for batch profiling - Color-coded warnings for slow hooks (>5 seconds) and output size limits - Specialized UserPromptSubmit limit checking for hooks that inject context into Claude prompts - Comprehensive test coverage with 493 test cases for profiling system validation - **Enhanced Hook Testing Infrastructure**: Improved hook validation and testing capabilities - Added intelligent test payload generation for different hook types (PostToolUse, Stop, UserPromptSubmit) - Enhanced file-guard hook testing with support for all AI ignore file formats (`.agentignore`, `.aiignore`, `.aiexclude`, `.geminiignore`, `.codeiumignore`, `.cursorignore`) - Improved pattern matching for synthetic test file generation from glob patterns - Added comprehensive test transcript creation for hooks requiring conversation context - **Hook Performance Constants**: Standardized performance thresholds and limits - Added performance threshold constants for execution time and output size monitoring - Claude Code specific output limits (10k character truncation threshold) - Token estimation utilities for output size analysis ### Fixed - **File-guard Pattern Matching**: Improved ignore file processing and pattern handling - Fixed file-guard to check all supported ignore files (6 different AI tool formats) - Enhanced pattern-to-filepath conversion for more accurate testing - Improved glob pattern handling for nested directories and wildcards - **Hook Output Capture**: Enhanced hook execution output collection - Fixed hook profiling to properly capture both stdout and stderr streams - Improved output size limiting with 10MB buffer for large hook outputs - Enhanced error handling for hook execution failures during profiling - **Test Infrastructure**: Resolved testing framework issues - Fixed Date constructor mocking in tests to avoid breaking Date functionality - Improved mock fs-extra implementation to use proper Date constructors - Enhanced test stability and reliability across different execution environments ### Changed - **Hook Runner Architecture**: Enhanced hook execution and output management - Updated HookRunner with standalone hook execution function for profiling integration - Improved output capture system with memory limits and truncation handling - Enhanced hook execution error handling and graceful failure management - **Hook CLI Interface**: Extended command-line interface capabilities - Added profile subcommand to hooks CLI with iterations parameter support - Enhanced argument validation for profile command parameters - Improved CLI help text and command structure for profiling features ## [0.8.1] - 2025-08-24 ### Fixed - **Agent Registry Consistency**: Fixed agent registry to use consistent naming conventions - Updated testing agent reference from `vitest-expert` to `vitest-testing-expert` in agent registry - Ensures proper agent resolution and availability detection in spec:execute workflows - Aligns with standardized naming convention where framework-specific testing experts use `-testing-` suffix ### Added - **Iterative Code Review Workflow**: Enhanced spec:execute command with comprehensive quality assurance - Added Phase 2: Test Writing with automatic selection of appropriate testing experts - Added Phase 3: Code Review with code-review-expert integration for comprehensive analysis - Added Phase 4: Iterative Improvement with specialist agent selection for fixing identified issues - Added Phase 5: Atomic commit creation with standardized commit message format - Added Phase 6: Progress tracking with quality metrics monitoring - **Dynamic Agent Availability Detection**: Improved agent orchestration in spec:execute - Added real-time agent availability checks using `claudekit list agents` - Dynamic selection of specialized agents based on actual availability rather than hardcoded lists - Fallback to general-purpose agents when specialized experts are unavailable - Enhanced agent matching logic for optimal task assignment ### Changed - **Spec Execution Architecture**: Restructured spec:execute for better quality and maintainability - Reorganized workflow into 6 distinct phases with clear quality gates - Enhanced task management integration with both STM and TodoWrite systems - Improved error handling and blocked task management - Added comprehensive success criteria validation - Enhanced progress tracking with status monitoring and quality metrics ## [0.8.0] - 2025-08-24 ### Added - **Thinking Level Hook**: New AI reasoning enhancement hook - Added `thinking-level` hook that injects thinking keywords based on configuration - Supports levels 0-4 with keywords: '', 'think', 'think hard', 'think harder', 'ultrathink' - Configurable via `.claudekit/config.json` with `enabled` and `level` settings - Triggers on UserPromptSubmit events to enhance AI reasoning capabilities - **Enhanced Hook Event Support**: Extended hook system with new event types - Added SessionStart event support to codebase-map hook - Updated codebase-map hook to trigger on both SessionStart and UserPromptSubmit events - Added thinking-level hook configuration to setup command initialization - **JSON Response System for Hooks**: New structured hook output format - Updated UserPromptSubmit hooks to return JSON responses with additionalContext - Implemented character limit handling (9000 chars) for UserPromptSubmit context - Enhanced hook output with hookEventName tracking for better event handling - **cclint Configuration**: Added code linting configuration for Claude Code projects - Added `cclint.config.js` with extended schemas for claudekit-specific features - Support for agent categories, display names, and bundled subagents - Enhanced command and settings schema validation with claudekit extensions ### Changed - **Codebase Map Hook Behavior**: Improved context delivery and event handling - Updated codebase-map hook to support both SessionStart and UserPromptSubmit triggers - Enhanced hook description to reflect dual event support - Improved character limit handling with truncation for UserPromptSubmit events - **Hook Configuration Structure**: Enhanced claudekit configuration schema - Added thinking-level and thinking-budget configuration schemas - Extended hook configuration types with new validation rules - Updated setup command to include thinking-level in session initialization group - **Test Infrastructure**: Updated test expectations for new JSON response format - Modified hook tests to validate JSON response structure instead of console output - Updated integration tests to check hookSpecificOutput.additionalContext fields - Enhanced test coverage for new hook response patterns ### Fixed - **UserPromptSubmit Hook Output**: Standardized hook response format - Fixed codebase-map hook to use JSON responses instead of console.log output - Resolved hook output consistency issues with structured JSON responses - Fixed character limit enforcement for UserPromptSubmit context injection ## [0.7.3] - 2025-08-24 ### Added - **Codebase Map Automatic Updates**: New hook for keeping codebase maps synchronized - Added `codebase-map-update` hook that automatically refreshes codebase maps when files are modified - Enabled by default in project `.claude/settings.json` for Write|Edit|MultiEdit operations - Ensures AI assistants always have up-to-date project structure information - **Node.js Import Migration Script**: Automated tool for modernizing import statements - Added `scripts/migrate-node-imports.js` for converting legacy Node.js imports to `node:` prefix format - Supports comprehensive migration of 23+ Node.js built-in modules (fs, path, crypto, etc.) - Handles ES6 imports, require statements, and dynamic imports - **Documentation Validation System**: Comprehensive docs accuracy checking - Added `scripts/validate-docs.sh` for systematic documentation validation against implementation - Color-coded output and structured validation reports - Checks command documentation, example accuracy, and configuration consistency - **Comprehensive Codebase Map Guide**: Detailed documentation for codebase map functionality - Added `docs/guides/codebase-map.md` with complete setup and configuration instructions - Includes format explanations, filtering options, and performance optimization tips - Documents both manual and automatic codebase map management workflows - **Unused Parameters Detection**: New code quality hook - Added `check-unused-parameters` hook to Code Quality group in setup - Automatically detects and flags unused function/method parameters - Helps maintain cleaner, more maintainable code ### Fixed - **Code Search Agent Color Configuration**: Updated agent configuration for better UI consistency - Fixed `code-search` agent color from invalid "amber" to valid "purple" for Claude Code compatibility - Ensures proper agent display and color coding in the interface - **Node.js Import Standardization**: Modernized all Node.js import statements across the codebase - Standardized 65+ files to use `node:` prefix for all Node.js built-in modules - Updated imports for fs, path, crypto, child_process, os, and other built-ins - Improves bundling compatibility and prevents module resolution issues - Aligns with modern Node.js best practices and esbuild requirements ### Changed - **Documentation Structure Reorganization**: Improved documentation consistency and navigation - Restructured checkpoint guide from `checkpoint-workflow.md` to `checkpoint.md` to match other documentation format - Enhanced file-guard setup documentation with one-liner installation commands - Improved documentation organization for better user experience - **Setup Command Hook Groups**: Enhanced setup command with additional quality checks - Added `check-unused-parameters` to Code Quality hook group - Extended hook configuration with PreToolUse support for file-guard integration ### Documentation - Enhanced file-guard setup documentation with quick installation commands - Added comprehensive configuration examples and troubleshooting guides - Improved documentation structure consistency across all guides ## [0.7.2] - 2025-08-23 ### Added - **Code Search Agent**: New specialized codebase search agent with optimized file discovery capabilities - Added `code-search` agent for fast, parallel file pattern searches - Optimized for finding specific files, functions, or patterns across large codebases - Uses parallel tool execution for 3-10x faster search performance - Returns focused file lists without unnecessary exploration or analysis - Includes hook disabling configuration to prevent interference during search operations - **Subagent Hook Optimization System**: Intelligent hook management for improved performance - Added `subagent-detector.ts` utility for detecting subagent context from transcripts - Implemented `disableHooks` field in subagent schema to selectively disable hooks per agent - Added SubagentStop event detection in BaseHook for performance optimization - Enhanced subagent metadata parsing with frontmatter support for hook configuration - **Enhanced Hook Performance Profiling**: Improved hook execution monitoring and optimization - Extended BaseHook class with subagent detection capabilities - Added conditional hook execution based on subagent context - Integrated performance-aware hook skipping for specialized agents ### Fixed - **Vitest Process Management**: Resolved hanging test processes and worker accumulation issues - Added `getExecOptions` function with intelligent environment variable management - Implemented vitest-specific environment variables (VITEST_POOL_TIMEOUT, VITEST_POOL_FORKS, VITEST_WATCH) - Fixed test command execution to prevent orphaned vitest worker processes - Enhanced process cleanup with proper timeout and fork management - **Test Configuration Optimization**: Streamlined vitest configuration to prevent process hangs - Configured `singleFork: true` to run all tests in single child process - Disabled `fileParallelism` for predictable process management - Reduced `teardownTimeout` to 3000ms for faster test completion - Removed separate hook configuration file (vitest.hook.config.ts) to eliminate complexity - **Test Reliability Improvements**: Enhanced test cleanup and directory management - Added comprehensive test directory cleanup in beforeAll/afterAll hooks - Implemented retry logic for directory removal on Windows/locked file systems - Added unique test directory naming to prevent conflicts between concurrent test runs - Improved error handling and cleanup in list command tests - **Subagent Schema Validation**: Corrected field categorization and validation - Fixed `color` field classification as official Claude Code field (not Claudekit-specific) - Enhanced subagent frontmatter schema with proper field validation - Improved linter accuracy for subagent configuration files ### Changed - **Test Command Configuration**: Updated package.json test scripts for better performance - Removed `--reporter=dot` flag from `test:fast` command for clearer output - Streamlined test execution with consistent vitest configuration - **Hook Execution Model**: Enhanced conditional execution based on context - Modified BaseHook to support SubagentStop event handling - Improved hook performance by skipping unnecessary operations in subagent contexts - Updated self-review hook with better transcript path handling ## [0.7.1] - 2025-08-23 ### Added - **UserPromptSubmit Hook Support**: New trigger event for enhanced user interaction handling - Added `UserPromptSubmit` to supported trigger events in hook system - Extended hook configuration schema to support UserPromptSubmit event type - Added UserPromptSubmit hook arrays to project settings initialization ### Fixed - **Codebase Map Configuration**: Corrected default codebase map settings for better project navigation - Fixed include path from `src/**` to `cli/**` to properly target TypeScript implementation code - Changed default format from `auto` to `dsl` for more consistent code structure visualization - **Vitest Command Configuration**: Fixed test command execution flags - Added missing `run` flag to `test:watch` command (`vitest run --watch`) - Added missing `run` flag to `test:ui` command (`vitest run --ui`) - Ensures proper test execution behavior in watch and UI modes - **Command Template Embedding**: Enhanced create-command and create-subagent with embedded templates - Fixed template content to be embedded directly in command files instead of external references - Improved reliability of command creation by removing dependency on external template files - **Documentation Examples**: Corrected bash command syntax in command creation documentation - Fixed bash command examples in create-command documentation for proper shell execution - Updated command syntax examples to follow correct bash patterns ## [0.7.0] - 2025-08-22 ### Added - **Sensitive File Protection (PreToolUse Hook)**: New security hook that prevents AI access to sensitive files - Added `file-guard` hook that blocks access to `.env`, keys, credentials, SSH keys, and other sensitive files - Supports multiple AI tool ignore files: `.agentignore`, `.aiignore`, `.aiexclude`, `.geminiignore`, `.codeiumignore`, `.cursorignore` - Pattern merging from multiple ignore files with gitignore-style syntax support - Default protection patterns for common sensitive files when no ignore files exist - Negation pattern support (e.g., `.env*` but `!.env.example`) - Path traversal prevention and symlink resolution for comprehensive security - Added to CLI setup as "File Security" group with recommended installation - **Code Review Expert Integration**: Enhanced self-review hook with intelligent subagent detection - Added automatic detection of `code-review-expert` subagent availability - Dynamic suggestion to use code-review-expert when available in user directories - Enhanced self-review prompts with Task tool guidance for specialized code review - **Specification Validation Enhancement**: Improved overengineering detection in spec:validate command - Added aggressive overengineering detection patterns and YAGNI principle enforcement - Enhanced validation questions to identify premature optimization and feature creep - Added comprehensive patterns for detecting over-abstraction and unnecessary complexity ### Changed - **Hook Configuration**: Introduced PreToolUse event support in hook system - Added PreToolUse trigger event to base hook types and metadata - Updated hook runner to handle PreToolUse permission decisions with JSON responses - Enhanced hook payload structure to include `tool_name` for PreToolUse filtering - **CLI Setup**: Enhanced hook group configuration with new file security category - Added File Security group as recommended installation option - Improved hook group organization with PreToolUse event categorization ### Fixed - **PreToolUse Hook JSON Output**: Fixed JSON response format for file-guard hook - Corrected PreToolUse permission decision structure for Claude Code compatibility - Fixed hook response format to include proper `hookEventName` and decision fields - **Agent Loader**: Enhanced agent detection for user-installed vs embedded agents - Added `isAgentInstalledByUser` method to distinguish user-installed agents from embedded ones - Improved agent path resolution and existence checking - **File Guard Error Handling**: Improved error handling and setup integration - Enhanced ignore file parsing with better error recovery - Improved fallback to default patterns when ignore files fail to load ### Security - **Default File Protection**: Added comprehensive default patterns for sensitive file protection - Protects environment files (`.env`, `.env.*`) - Blocks access to private keys (`*.pem`, `*.key`) - Prevents access to cloud credentials (`.aws/credentials`) - Protects SSH keys (`.ssh/*`) - Blocks files outside project root via path traversal prevention ## [0.6.10] - 2025-08-21 ### Added - **Documentation Expert Subagent**: Comprehensive documentation specialist with advanced research capabilities - Added `documentation-expert` with expertise in structure, cohesion, flow, audience targeting, and information architecture - Specialized in detecting documentation anti-patterns and optimizing for user experience - Proactive usage for documentation quality issues, content organization, duplication, and readability concerns - Includes comprehensive research reports for documentation structure and cohesion analysis - **Markdown Link Validation**: Automated broken link detection for documentation quality assurance - Added `.markdown-link-check.json` configuration with pattern ignoring and replacement rules - Integrated link checking into release preparation process to halt releases with broken documentation links - Added GitHub-specific URL replacements and timeout/retry configurations for reliable link validation - Enhanced release script with comprehensive documentation link verification before release ### Changed - **Self-Review Hook Focus Areas**: Restructured self-review prompts with enhanced implementation completeness detection - Reordered focus areas to prioritize "Implementation Completeness" first, targeting mock implementations and placeholder code - Enhanced questions to detect incomplete implementations, hardcoded return values, and TODO placeholders in production code - Reorganized focus areas: Implementation Completeness โ†’ Testing โ†’ Code Quality โ†’ Integration & Refactoring โ†’ Codebase Consistency - Updated default configuration in both hook implementation and `.claudekit/config.json` example - **CLI Expert Release Workflow**: Enhanced GitHub release workflow with changelog content extraction - Improved release template to automatically extract and format changelog entries for release descriptions - Added intelligent changelog parsing to populate GitHub release notes with structured content ### Fixed - **Documentation Links**: Resolved all broken markdown documentation links across the codebase - Fixed broken internal links in installation guides, reference documentation, and examples - Updated outdated GitHub repository URLs and internal reference paths - Corrected links in hook configuration examples and README files - Added systematic link validation to prevent future documentation link rot ## [0.6.9] - 2025-08-21 ### Changed - **Command Namespace Standardization**: Renamed AI assistant command namespace for consistency - Renamed command namespace from `agent-md` to `agents-md` for consistency with filename conventions - Updated all command references in CLI setup configuration from `agent-md:*` to `agents-md:*` - Updated documentation references from `AGENT.md` to `AGENTS.md` throughout the codebase - Updated lint error messages to reference `AGENTS.md/CLAUDE.md` for command lookup - Maintains full backward compatibility - existing functionality unchanged ## [0.6.8] - 2025-08-20 ### Changed - **CLI Command Naming**: Renamed `validate` command to `doctor` for better CLI conventions and clarity - Updated CLI registration from `claudekit validate` to `claudekit doctor` - Renamed command implementation file from `validate.ts` to `doctor.ts` - Updated all documentation references and setup guidance to use `doctor` terminology - Maintains backward compatibility with same functionality and options ### Added - **Codebase Map Filtering Configuration**: Enhanced codebase-map hook with advanced filtering capabilities - Added `include` and `exclude` pattern support in `.claudekit/config.json` configuration - Added support for glob patterns to filter which files are included in codebase maps - Added filtering to codebase-map format command with `--include` and `--exclude` flags - Enhanced configuration schema with validation for include/exclude string arrays - Improved TypeScript types for CodebaseMapConfig with new filtering options ### Fixed - **Test Suite**: Updated test expectations to reflect new diagnostic terminology - Updated all test files to use `doctor` command instead of `validate` - Fixed test assertions to match new command behavior and output - Updated integration tests for proper workflow validation with new command name ## [0.6.7] - 2025-08-20 ### Added - **Subagent Tools Field Validation**: Enhanced subagent linter with comprehensive empty tools field detection - Added detection for empty `tools` fields (null, empty string, whitespace-only) in subagent configurations - Added warning messages explaining that empty tools fields grant NO tools to subagents - Added guidance to either remove the field entirely (inherits all tools) or specify tools explicitly - Added comprehensive test coverage with 12 test cases for various empty field scenarios - Enhanced validation logic to handle YAML null values, comments-only fields, and edge cases ### Fixed - **Code Review Command**: Fixed YAML parsing error in `/code-review` command - Properly quoted `argument-hint` field to prevent YAML parsing failures - Ensures command can be properly loaded and executed without syntax errors ## [0.6.6] - 2025-08-20 ### Fixed - **Test Infrastructure**: Improved test script reliability and path resolution - Fixed path resolution in `tests/unit/test-subagents.sh` to use absolute paths with proper script directory detection - Updated agent file reference from `expert.md` to `typescript-expert.md` to match actual file structure - Enhanced test failure handling to treat validation warnings as non-blocking for CI stability - Improved script portability using `${BASH_SOURCE[0]}` instead of `$0` for more reliable path detection ## [0.6.5] - 2025-08-20 ### Fixed - **ESM/CommonJS Compatibility**: Fixed module compatibility issues for better cross-platform support - Fixed import.meta.url usage with fallback to __filename for CommonJS builds - Updated main entry point detection to work in both ESM and CommonJS environments - Enhanced module loading with proper fallbacks for Node.js built-in patterns - Improved CLI startup reliability across different Node.js module systems - **Dynamic Import Integration**: Converted oh-my-logo to dynamic import for ESM compatibility - Changed from static import to dynamic import in setup command to prevent bundling issues - Added graceful fallback when oh-my-logo cannot be loaded - Maintains ASCII art functionality while improving module compatibility ### Changed - **Build System**: Enhanced build configuration for better Ubuntu compatibility and dependency management - Changed build output format from ESM to CommonJS for improved compatibility - Updated main package.json entry points to use .cjs extensions - Enhanced esbuild configuration with better external dependency handling - Added comprehensive dependency validation scripts for build verification ### Added - **Enhanced Validation Infrastructure**: Expanded build and dependency validation capabilities - Enhanced `scripts/validate-dependencies.js` with improved bundling strategy validation - Enhanced `scripts/test-clean-install.sh` with comprehensive CLI functionality testing - Added support for validating both bundled and external dependencies - Improved error reporting and validation feedback for development workflows ## [0.6.4] - 2025-08-19 ### Added - **Production Validation Scripts**: Added comprehensive build and installation validation - Added `scripts/validate-dependencies.js` for runtime dependency validation - Added `scripts/test-clean-install.sh` for clean environment testing - Added `validate:deps`, `validate:build`, `validate:install`, and `prepublish:validate` npm scripts - Validates external dependencies are properly declared in package.json - Tests CLI functionality in isolated environment to catch missing dependencies ### Changed - **Build Configuration**: Enhanced ESBuild configuration with explicit external dependencies - Updated build scripts to properly externalize `fs-extra`, `@inquirer/prompts`, `commander`, `ora`, and `oh-my-logo` - Improved bundle optimization by separating external vs bundled dependencies - **ESLint Configuration**: Enhanced JavaScript file linting with comprehensive rules and globals - Added dedicated JavaScript file configuration with Node.js globals - Added code quality rules for modern JavaScript development - Improved linting coverage for CLI scripts and configuration files ### Fixed - **CLI Startup Robustness**: Improved CLI initialization error handling - Changed show commands registration failure from error to debug logging - CLI no longer fails to start when show commands cannot be registered - Enhanced graceful degradation for optional command modules ## [0.6.3] - 2025-08-19 ### Changed - **Command Naming**: Renamed `/review` command to `/code-review` to avoid conflicts with native Claude Code commands - Updated command name in all documentation references (README.md, docs/) - Updated symlink structure and agent references - Maintains all existing functionality with improved naming consistency ## [0.6.2] - 2025-08-19 ### Added - **Fast Test Configuration**: Added optimized Vitest configuration for hook execution - Added `vitest.hook.config.ts` with ultra-fast settings optimized for Stop hook timeout constraints - Added `test:fast` npm script for rapid testing without coverage - Added `test-project` command to claudekit configuration for project-level test execution - Enhanced test infrastructure with comprehensive configuration validation tests ### Changed - **Agent MD Init Enhancement**: Improved `/agents-md:init` command to encourage delegation to specialized agents - Enhanced command to emphasize the importance of using domain-specific subagents - Updated guidance to promote proactive delegation patterns for better task management - Improved documentation structure for specialized agent usage ### Fixed - **TypeScript Import Issues**: Resolved fs-extra import compatibility across hook system - Fixed Node.js module imports from namespace to default imports in `cli/hooks/base.ts` - Fixed imports in `cli/hooks/runner.ts` and `cli/hooks/utils.ts` for better ES module compatibility - Updated test files to use consistent import patterns - Ensures compatibility with modern Node.js module resolution ## [0.6.1] - 2025-08-19 ### Added - **Triage Expert Subagent**: New context gathering and problem diagnosis specialist - Added `triage-expert` for initial error analysis and routing to specialized experts - Universal agent for handling errors, performance issues, or unexpected behavior - Provides comprehensive diagnostic context before delegating to domain experts - Includes strict diagnosis boundaries and mandatory cleanup of temporary debug code - **Specialized Subagent Integration**: Enhanced workflow commands with intelligent delegation - Added subagent integration patterns to `spec:create`, `spec:validate`, and `validate-and-fix` commands - Commands now intelligently delegate to domain experts based on task requirements - Documented specialized subagent usage patterns in creating-commands guide - **Mandatory Subagent Delegation Guidelines**: Enhanced AGENTS.md with strict delegation requirements - Added mandatory requirement to use specialized subagents for all technical issues - Documented clear routing patterns for different types of problems - Prevents direct problem solving without specialist expertise ### Changed - **Agent Naming Standardization**: Renamed `code-reviewer` to `code-review-expert` for consistency - **Agent Description Format**: Fixed all agent descriptions from first-person to second-person format - Updated 12+ expert agents with consistent "You are" description pattern - Improves clarity and consistency across all subagents ### Fixed - **TypeScript Import Issues**: Resolved module import and compilation problems - Fixed Node.js module imports from default to namespace imports across all files - Added `downlevelIteration` to tsconfig.json to resolve iterator issues - Applied consistent ESLint/Prettier formatting throughout codebase - Ensured all 682 tests pass with improved validation logic - **CLI Test Infrastructure**: Added timeout mechanism to prevent hanging tests - Fixed infinite hanging in "should handle missing .claude directory gracefully" test - Added configurable timeout parameter (default 5s, 10s for complex operations) - Added proper process cleanup and error handling for spawned CLI commands - Improved test reliability with SIGTERM handling and timeout controls ## [0.6.0] - 2025-08-18 ### Added - **Codebase Map Integration**: Added comprehensive project structure mapping using external codebase-map CLI - Added `CodebaseMapHook` (UserPromptSubmit) providing invisible codebase context on first prompt per session - Added `CodebaseMapUpdateHook` (PostToolUse) for incrementally updating maps when files change - Automatically scans and indexes project structure with AST-based analysis - Generates optimized output format for LLM consumption with dependency tracking - Session tracking prevents duplicate context loading within same Claude Code session - Automatic cleanup of old session files (older than 7 days) - Requires external [codebase-map](https://github.com/carlrannaberg/codebase-map) CLI installation - **Shared Hook Utilities**: Added reusable utilities for hook development - Added `codebase-map-utils.ts` for shared codebase map generation functionality - Added `SessionTracker` class in `session-utils.ts` for per-session state management - Added comprehensive file modification time tracking and session flag management - **Enhanced Git Commands**: Added git-expert integration capabilities to git workflow commands - Enhanced `/git:commit`, `/git:status`, and `/git:push` commands with Task tool support - Added recommendations for using git-expert subagent for complex scenarios - **Enhanced Hook Registry**: Added support for UserPromptSubmit and SubagentStop event hooks - Updated hook registry to support new event types alongside PostToolUse and Stop - Enhanced hook runner with proper event type handling and metadata processing - **Refactoring Expert Subagent**: Added comprehensive code smell detection and refactoring guidance - Added `refactoring-expert` subagent with 25+ code smell detection patterns - Provides systematic refactoring guidance with before/after examples - Includes safe refactoring process with test-driven incremental changes - Covers 6 categories: Composing Methods, Moving Features, Organizing Data, Simplifying Conditionals, Making Method Calls Simpler, and Dealing with Generalization ### Changed - **Dynamic Agent Count Display**: Fixed setup command to show actual number of available agents instead of hardcoded count - Setup command now dynamically calculates and displays the correct number of agents in the "Install All" option - Eliminates maintenance burden of manually updating agent counts when new agents are added - Improves accuracy and prevents misleading information in the setup interface - **Hook List Command**: Enhanced hooks listing to show actual configured hooks from project settings - Added `listHooks()` function that processes both project and user configurations - Displays hook source (project/user), associated events, and configured matchers - Improved hook name extraction from embedded system commands - **Enhanced Test Coverage**: Added comprehensive test suites for new functionality - Added `tests/unit/codebase-map.test.ts` with 30+ test cases for codebase mapping - Added `tests/integration/codebase-hooks.test.ts` for end-to-end hook testing - Added `tests/unit/list-hooks.test.ts` with extensive hook listing validation - Enhanced test helpers and utilities for better test reliability ### Fixed - **ESLint Configuration**: Updated ESLint configuration to support TypeScript 6.0+ and modern patterns - Enhanced parser options for better TypeScript compatibility - Updated rule configurations for cleaner code standards - **Test Suite Reliability**: Improved test suite stability and reduced flaky tests - Enhanced debounce boundary tests with better timing control - Improved mock handling for filesystem operations and external command execution - Fixed test isolation issues with better cleanup and setup patterns ## [0.5.0] - 2025-08-17 ### Added - **Comprehensive Code Review System**: Added multi-aspect code review functionality with parallel agent processing - Added `/code-review` command for intelligent code reviews using specialized parallel agents - Added `code-review-expert` agent with 6 focused review aspects: architecture & design, code quality, security & dependencies, performance & scalability, testing quality, and documentation & API design - Added structured report format with issue prioritization (Critical/High/Medium), type icons (๐Ÿ”’ Security, ๐Ÿ—๏ธ Architecture, โšก Performance, ๐Ÿงช Testing, ๐Ÿ“ Documentation, ๐Ÿ’ฅ Breaking), and actionable feedback - Added intelligent agent selection based on file types and review scope (documentation-only, test files, config files, or full source code) - Added quality metrics scoring system and issue distribution tracking for comprehensive analysis - **Enhanced Expert Agent Collection**: Added comprehensive domain-specific expert agents with enhanced capabilities - Added `ai-sdk-expert` for Vercel AI SDK v5 development, streaming, and model integration - Added `cli-expert` for npm package CLI development with Unix philosophy and argument parsing - Added `nestjs-expert` for Nest.js framework development with dependency injection and testing - Enhanced multiple existing agents with comprehensive code review checklists and quality validation capabilities - Added specialized build tool experts: `build-tools-vite-expert` and `build-tools-webpack-expert` - Added infrastructure experts: `infrastructure-docker-expert` and `infrastructure-github-actions-expert` - Added frontend experts: `frontend-css-styling-expert` and `frontend-accessibility-expert` - Added comprehensive testing suite: `testing-expert`, `jest-testing-expert`, `vitest-testing-expert`, and `e2e-playwright-expert` - Added database experts: `database-expert`, `database-postgres-expert`, and `database-mongodb-expert` - Added additional specialized experts: `git-expert`, `nodejs-expert`, `react-expert`, `react-performance-expert`, `typescript-expert`, `typescript-build-expert`, `typescript-type-expert`, and `code-quality-linting-expert` - **Developer Documentation**: Added comprehensive guides for subagent and command development - Added `docs/guides/creating-subagents.md` with detailed instructions for creating domain expert subagents - Added `docs/guides/creating-commands.md` with command development patterns and best practices ### Changed - **Agent Architecture**: Upgraded review system to use 6 focused agents architecture for comprehensive parallel code analysis - **Review Efficiency**: Enhanced review process with intelligent agent selection based on file types and scope to reduce unnecessary analysis overhead - **Report Structure**: Improved review report format with consistent type icons, priority levels, and actionable feedback structure ### Fixed - **Review Template Organization**: Moved review instructions outside of report template to clarify systemic issues section and improve template readability - **Agent Selection Logic**: Enhanced review efficiency by launching only relevant agents based on the specific files and changes being reviewed ## [0.4.0] - 2025-08-16 ### Added - **CLI Show Command**: New `claudekit show` command for exposing agent and command prompts in headless mode - Added `claudekit show agent ` subcommand to display agent prompts with support for text and JSON output formats (--format text|json) - Added `claudekit show command ` subcommand to display command prompts with support for text and JSON output formats (--format text|json) - Added comprehensive loader infrastructure with `AgentLoader` and `CommandLoader` classes for robust file resolution - Added support for multiple agent/command resolution strategies including direct file matching, category/name patterns, and frontmatter name field matching - Added proper error handling with helpful suggestions to use `claudekit list` commands when items are not found - **Advanced Agent Discovery**: Enhanced agent resolution with intelligent path matching and frontmatter parsing - Added support for category-based agent organization (e.g., `typescript/expert`) - Added automatic `-expert` suffix handling for simplified agent references - Added recursive frontmatter name field matching for flexible agent identification - Added robust error handling for malformed frontmatter with graceful fallbacks - **Command Resolution System**: Enhanced command discovery with namespace support and flexible path resolution - Added support for namespaced commands using colon syntax (e.g., `spec:create` โ†’ `spec/create.md`) - Added recursive directory traversal for commands in any subdirectory structure - Added intelligent allowed-tools parsing supporting both string and array formats from frontmatter - **Specialized AI Expert Subagents**: Added comprehensive collection of domain-specific AI subagents - Added `ai-sdk-expert` for Vercel AI SDK v5 development, streaming, and model integration - Added `cli-expert` for npm package CLI development with Unix philosophy and argument parsing - Added `nestjs-expert` for Nest.js framework development with dependency injection and testing ### Changed - **CLI Architecture**: Refactored CLI initialization to support dynamic command registration - Modified `runCli()` function to be async and support dynamic import of show commands - Enhanced CLI error handling with proper async error propagation and graceful failure modes - Updated command registration pattern to support modular command loading - **Agent List Display**: Enhanced agent listing with frontmatter-based filtering and display names - Modified agent filtering to use frontmatter `name` field instead of filename for more accurate matching - Updated display logic to show human-readable names from frontmatter rather than technical filenames - Improved token estimation and frontmatter extraction for better performance and accuracy ### Fixed - **Test Output**: Configured vitest for minimal output to reduce context pollution during test runs - **NPM Notices**: Filtered npm notices from test stderr output to prevent test noise - **Model References**: Updated AI model references to current August 2025 versions with accurate context sizes ## [0.3.11] - 2025-08-16 ### Added - **Agent Listing Support**: Enhanced the `claudekit list` command with comprehensive agent discovery and listing functionality - Added `agents` as a new valid list type alongside `hooks`, `commands`, and `config` - Added `listAgents()` function that recursively scans `.claude/agents/` directory for agent files - Added token count estimation for agents using a heuristic of ~1 token per 4 characters - Added agent categorization based on directory structure (e.g., `general`, `typescript`, `react`) - Added frontmatter parsing to extract agent descriptions from YAML metadata - Enhanced display with grouped agent output showing category organization and token counts - Updated `AgentInfo` interface with `category`, `tokens`, and enhanced metadata fields - Modified command validation to include agents type checking with proper TypeScript discrimination ### Changed - **Token Count Integration**: Added token estimation and display across commands and agents - Enhanced `CommandInfo` interface to include `tokens` field for command complexity measurement - Added `estimateTokens()` and `formatTokens()` utility functions for consistent token display - Updated command listing display to show token counts instead of file sizes for better relevance - Modified display formatting to show token counts in human-readable format (e.g., "1.2k tokens") - **List Command Type Validation**: Updated valid types from `['all', 'hooks', 'commands', 'settings', 'config']` to `['all', 'hooks', 'commands', 'agents', 'config']` - Removed deprecated `settings` type in favor of standardized `config` type naming - Enhanced type discrimination in result processing to properly distinguish between commands and agents ## [0.3.10] - 2025-08-15 ### Added - **SubagentStop Hook Support**: Added support for hooks to trigger when subagents complete their tasks - Added `SubagentStop` to the `HookEvent` type alongside existing `PostToolUse` and `Stop` events - Enhanced hook metadata to support arrays of trigger events for multi-event hooks - Updated all completion validation hooks (`typecheck-project`, `lint-project`, `test-project`, `check-todos`, `self-review`) to trigger on both `Stop` and `SubagentStop` events - Added `SubagentStop` configuration support in the setup process with proper hook grouping and merging - Enables quality validation and checkpointing when subagent tasks complete, ensuring consistency across all Claude Code workflows - **Show Command Specification**: Added comprehensive specification for exposing agent and command prompts in headless mode - Added CLI expert subagent specification for advanced command-line interface development guidance - Enhanced specification creation and validation workflows for better development planning ### Changed - **Hook Trigger Configuration**: Enhanced hook setup to support multiple trigger events per hook - Modified hook metadata system to accept either single trigger events or arrays of events - Updated setup process to properly configure hooks for multiple trigger points - Improved hook merging logic to handle `SubagentStop` configurations alongside existing events ## [0.3.9] - 2025-08-14 ### Fixed - **Comment Replacement Hook File Path Integration**: Enhanced the check-comment-replacement hook to use file path information for smarter validation - Added file path extraction for both Edit and MultiEdit operations to enable file-type-aware validation - Added automatic exclusion of documentation files (`.md`, `.mdx`, `.txt`, `.rst`) from comment replacement validation - Enhanced debug logging with detailed context payload information when DEBUG environment variable is set - Improved validation accuracy by skipping comment replacement checks on documentation files where such patterns are legitimate - Reduced false positives for documentation edits while maintaining code quality enforcement for actual source files ## [0.3.8] - 2025-08-14 ### Changed - **Doctor Command Flag Standardization**: Replaced `--detailed` flag with universal `--verbose` flag for consistency - Removed the `--detailed` option from the doctor command CLI interface - Updated validation output logic to use `--verbose` instead of `--detailed` for enhanced details - Changed output header from "Detailed Validation:" to "Validation Details:" for clarity - Maintains same functionality while providing consistent flag naming across the CLI ### Fixed - **Comment Replacement Detection**: Enhanced the check-comment-replacement hook to reduce false positives - Improved hash comment pattern to exclude markdown headers (e.g., `##`) from comment detection - Enhanced block comment continuation pattern to require space after `*` for more precise matching - Added size difference analysis to distinguish between content replacement and content deletion - Improved replacement detection logic to avoid flagging legitimate content removal as violations - Reduced false positives when users delete sections of code rather than replacing them with comments ## [0.3.7] - 2025-08-14 ### Added - **Agent Discovery Documentation**: Enhanced `/agents-md:init` command with comprehensive subagent documentation - Added automatic discovery of available subagents in the project - Integrated detailed subagent usage guidelines into generated AGENTS.md files - Included proactive delegation patterns and examples for 24+ specialized agents across 7 categories - Added when-to-use guidance for domain-specific tasks (React performance, TypeScript errors, build optimization, etc.) ### Fixed - **Self-Review Duplicate Detection**: Improved reliability of self-review hook duplicate prevention - Enhanced transcript parsing to properly detect Stop hook output in tool_result content - Fixed marker detection to check both user message content and toolUseResult.reason fields - Improved parsing of tool results where self-review markers are embedded in JSON responses - Added debug logging for self-review trigger analysis to aid troubleshooting - **TranscriptParser Tool Result Handling**: Enhanced transcript analysis for better hook integration - Added support for parsing tool_result type content blocks in user messages - Improved detection of Stop hook feedback that appears in Claude Code UI - Better handling of structured JSON content within tool results - Enhanced marker finding logic to work reliably with embedded hook system output ### Changed - **Self-Review Configuration**: Updated example configuration to reflect current best practices - Simplified configuration example with focus on essential settings (timeout, targetPatterns, focusAreas) - Updated example targetPatterns to show TypeScript/JavaScript file filtering - Enhanced focusAreas example configuration with clear structure and purpose ## [0.3.6] - 2025-08-14 ### Changed - **Self-Review Hook Configuration**: Simplified the self-review hook by removing complexity and focusing on essential functionality - Removed `triggerProbability` configuration option - hook now triggers deterministically when code changes are detected - Removed `messageWindow` configuration parameter - replaced with intelligent change detection since last review - Enhanced change detection logic to check for new file modifications since the last self-review marker - Added default 200-entry lookback limit when no previous review marker exists to prevent excessive history scanning - Improved transcript parsing logic for more reliable file change detection ### Fixed - **TranscriptParser Logic**: Enhanced transcript analysis methods for better change detection - Added `hasFileChangesInRange()` method for checking changes within specific entry ranges - Improved `hasFileChangesSinceMarker()` to properly handle cases where no previous marker exists - Fixed file change detection to be more precise and reduce false positives - **Self-Review Configuration Schema**: Updated TypeScript configuration schema to remove deprecated options - Removed `triggerProbability` and `messageWindow` from `SelfReviewConfigSchema` - Streamlined configuration interface to focus on essential settings (timeout, targetPatterns, focusAreas) - **Test Suite Updates**: Updated test cases to reflect simplified configuration and behavior - Removed tests for probability-based triggering and message window configuration - Enhanced tests for new change detection logic and marker-based tracking - Improved test coverage for transcript parsing edge cases ## [0.3.5] - 2025-08-13 ### Added - **TranscriptParser Utility**: New comprehensive transcript analysis system for Claude Code session parsing - Added `cli/utils/transcript-parser.ts` with intelligent conversation message grouping - Supports complex transcript parsing with UI message detection and tool use analysis - File pattern matching with glob support and negative patterns (e.g., `['**/*.ts', '!**/*.test.ts']`) - Smart message windowing that matches Claude Code UI behavior (dots in conversation view) - Functions for detecting file changes, finding markers, and analyzing recent activities - **Self-Review Hook Enhancements**: Major improvements to the self-review system - Added configurable `messageWindow` setting (default: 15 UI messages) - Added `targetPatterns` configuration for glob-based file filtering - Enhanced `focusAreas` configuration with custom question sets - Intelligent duplicate prevention with marker-based tracking (`๐Ÿ“‹ **Self-Review**`) - Smart file change detection that respects time windows and file patterns - **Comprehensive Test Suite**: Added extensive test coverage for new functionality - `tests/unit/transcript-parser.test.ts` with 25+ test cases covering all parser features - `tests/unit/transcript-parser-grouping.test.ts` with real-world transcript parsing scenarios - `tests/unit/self-review.test.ts` with comprehensive hook behavior validation - Tests cover message grouping, glob patterns, probability triggers, and configuration scenarios - **Configuration Examples**: Added `examples/settings.self-review.json` with documented configuration options ### Changed - **Self-Review Hook Architecture**: Complete rewrite using the new TranscriptParser system - Replaced manual conversation parsing with robust TranscriptParser utility - Improved file change detection accuracy with configurable target patterns - Enhanced message windowing to match Claude Code UI behavior exactly - Simplified hook logic by delegating transcript analysis to dedicated parser - Better duplicate prevention using consistent marker detection - **TypeScript Hook Error Messages**: Enhanced error feedback with specific command information - TypeScript hooks now include the exact command (e.g., `npm run typecheck`) in error messages - Updated `formatTypeScriptErrors()` utility to accept optional command parameter - More actionable error messages that tell users exactly what to run for verification - **Hook Utilities**: Enhanced `check-todos` hook to use new TranscriptParser system - Replaced manual transcript parsing with centralized parser utility - Improved reliability and consistency across transcript-reading hooks ### Fixed - **Message Counting Accuracy**: Fixed UI message counting to match Claude Code interface - Messages are now grouped correctly (text + tools = one UI message) - Accurate detection of when assistant messages get dots in the conversation view - Proper handling of standalone tool-only messages (like TodoWrite) - **Self-Review Trigger Logic**: Improved trigger conditions and duplicate detection - Fixed probability-based triggering with proper random number generation - Enhanced file pattern matching to avoid false positives on documentation files - Corrected transcript path handling and validation - Better debug logging for troubleshooting trigger behavior - **Test Infrastructure**: Improved test reliability and coverage - Fixed vitest reporter configuration (switched from 'verbose' to 'default') - Enhanced mock system for filesystem operations and transcript parsing - Better test assertions that actually validate functionality rather than side effects ### Security - **File Pattern Validation**: Enhanced glob pattern validation to prevent path traversal issues - **Transcript Access**: Secure transcript file access with proper path expansion and validation ## [0.3.4] - 2025-08-13 ### Added - **Claudekit Configuration System**: New centralized configuration system with `.claudekit/config.json` support - Added `cli/utils/claudekit-config.ts` with `loadClaudekitConfig()` and `getHookConfig()` functions - Supports JSON schema validation for configuration files - Enables project-specific hook configuration with fallback to sensible defaults - Added configuration example file at `.claudekit/config.json.example` ### Changed - **Hook Configuration Architecture**: Standardized all hooks to use the new configuration system - All hooks now use `getHookConfig()` instead of accessing `this.config` directly - Improved type safety with proper TypeScript interfaces for each hook's configuration - Configuration loading with graceful fallback when config files are missing or invalid - Updated hooks: `create-checkpoint`, `lint-changed`, `lint-project`, `test-changed`, `test-project`, `typecheck-changed`, `typecheck-project`, `self-review` - **Self-Review Hook Improvements**: Simplified and enhanced the self-review hook functionality - Replaced persona-based system with structured focus areas (Refactoring & Integration, Code Quality, Consistency & Completeness) - Added configurable trigger probability (defaults to 70%, configurable via `triggerProbability` in config) - Improved question selection with one question per focus area for better coverage - Streamlined message templates with clearer, more actionable feedback - Enhanced configuration support for timeout and trigger probability settings ### Fixed - **Configuration Loading**: Robust configuration loading with proper error handling and debug logging - **Hook Parameter Interfaces**: All hooks now have properly typed configuration interfaces - **Test Suite Updates**: Updated all hook tests to mock the new configuration system properly ## [0.3.3] - 2025-08-13 ### Added - **Self Review Hook**: New validation hook that prompts critical self-review with randomized senior developer personas and questions - Triggers on Stop events after code file modifications with 70% probability - Features 5 senior developer personas with distinct styles and catchphrases - Includes 3 review frameworks focusing on code coherence, integration, and overall health - Smart detection of code file changes vs documentation/config files - Analyzes conversation history to only trigger when actual code files were modified - Supports 20+ programming languages (.ts, .tsx, .js, .jsx, .py, .java, .cpp, .go, .rs, .swift, .kt, .rb, .php, .scala, .vue, .svelte, etc.) - Prevents hook loops and provides structured feedback with randomized question selection ### Fixed - **Self Review Hook Targeting**: Improved hook to only trigger on actual code file changes, ignoring documentation and configuration files - Excludes README, CHANGELOG, LICENSE, .md, .txt, .json, .yaml, .yml, .toml, .ini, .env, .gitignore, .dockerignore files - Uses conversation transcript analysis to detect recent code editing tool usage - Prevents false triggers on non-code modifications ## [0.3.2] - 2025-08-12 ### Added - **Check Unused Parameters Hook**: New validation hook that detects lazy refactoring where function parameters are prefixed with underscore instead of being properly removed - Analyzes Edit/MultiEdit operations to detect parameter name changes from `param` to `_param` - Provides detailed feedback on proper parameter handling practices - Supports TypeScript/JavaScript function declarations, arrow functions, methods, and constructors - Helps maintain clean function signatures by encouraging proper parameter removal ### Changed - **Comment Replacement Hook Scope**: Refined `check-comment-replacement` hook to only trigger on Edit/MultiEdit operations (excluding Write operations for better precision) ### Fixed - **Test Suite Robustness**: Updated component discovery tests to be more resilient to changes in embedded hook counts - Tests now use dynamic counts and functional assertions instead of hardcoded numbers - Improved test reliability for component registry validation - Better handling of embedded hook discovery in test scenarios ## [0.3.1] - 2025-08-12 ### Added - **Check Comment Replacement Hook**: New validation hook that detects when functional code is replaced with comments during edits, helping maintain clean codebases - **Symlinks Management Script**: New `npm run symlinks` command and `scripts/create-symlinks.sh` for creating/updating symlinks from `.claude/` to `src/` directories for development ### Changed - **Hook Registration System**: Simplified hook registration from 8 manual steps to just 2 steps using metadata-driven approach - Hooks now use a single source of truth pattern with automatic registry building - Settings generation automated from hook metadata via matcher field - Components discovery now fully automated - Eliminated 60+ line switch statement in favor of metadata-driven logic ### Fixed - **TypeScript 'any' Detection**: Improved `check-any-changed` hook to avoid false positives in strings, comments, and test utilities - Added `removeStringsAndComments()` method to strip content before validation - Uses dynamic regex construction to prevent self-detection - Preserves original line content in error messages while analyzing cleaned content - Handles single/double quotes, template literals, and both comment styles - **Code Formatting**: Applied prettier formatting across 12+ files to ensure consistent code style ### Security - Enhanced comment replacement detection to prevent code being hidden behind explanatory comments instead of clean deletion ## [0.3.0] - 2025-08-11 ### Added - **Domain Expert Subagents Suite**: Comprehensive library of 24+ specialized AI subagents across 7 categories - **Build Tools**: Vite Expert, Webpack Expert with advanced build optimization expertise - **Code Quality**: General linting expert with cross-language support - **Database**: MongoDB Expert, PostgreSQL Expert, and general database expert - **DevOps**: Docker Expert, GitHub Actions Expert, and general DevOps specialist - **Framework**: Next.js Expert with App Router and Server Components expertise - **Frontend**: Accessibility Expert (WCAG 2.1/2.2), CSS Styling Expert with modern CSS features - **Git**: Git Expert with merge conflicts, branching strategies, and repository management - **Node.js**: Node.js Expert with async patterns, performance optimization, and ecosystem knowledge - **React**: React Expert and React Performance Expert for optimization challenges - **Testing**: Jest Expert, Vitest Expert, and general testing framework specialist - **TypeScript**: TypeScript Expert, TypeScript Build Expert, and TypeScript Type Expert for advanced type system challenges - **Agent Selection System**: Intelligent categorized agent selection with radio group interface - **Non-Interactive Agent Installation**: `--agents` flag for automated CI/CD workflows (e.g., `--agents typescript-expert,react-expert`) - **Subagent Linting Tools**: `claudekit lint-subagents` command with frontmatter validation - **Command Linting Tools**: `claudekit lint-commands` command for slash command validation - **Agent Registry System**: Dynamic agent grouping and bundled agent selection - **Enhanced Setup Wizard**: Three-step selection process (Commands โ†’ Hooks โ†’ Agents) with improved UX - **MCP Tool Support**: Validation for Model Context Protocol tools in slash commands ### Changed - **Setup Command Architecture**: Complete redesign with grouped selection interface - **Agent Installation Flow**: Integrated agents into unified component system alongside commands and hooks - **CLI Interface**: Added `--skip-agents`, `--agents `, and improved flag handling - **Progress Reporting**: Enhanced progress indicators with agent count display - **Component Discovery**: Extended registry to handle agent categorization and bundling - **Validation System**: Improved frontmatter validation for both commands and subagents ### Fixed - **Frontmatter Issues**: Cleaned up all subagent frontmatter validation problems - **Linter Summaries**: Improved linter output formatting and removed redundant self-references - **Default Directory Handling**: Corrected default directories for lint commands - **Array Tools Field**: Graceful handling of array-based tools field in subagent linter - **Agent Validation**: Improved validation to properly check for required fields and complete symlinks - **Color Format Validation**: Added hex color format validation with helpful CSS color suggestions ### Security - **Tool Restrictions**: Enhanced allowed-tools validation across all new linting commands - **Frontmatter Validation**: Strict validation of subagent and command metadata to prevent malformed configurations ## [0.2.1] - 2025-08-07 ### Fixed - **Test Hook Timeout Handling**: Improved test-project hook with 55-second timeout limit and better error messages for Claude Code's 60-second hook timeout - **Vitest 3 Compatibility**: Updated test framework from vitest ^2.1.8 to ^3.2.4 and fixed test helper mocking compatibility - **Debug Logging Cleanup**: Removed unnecessary debug console output from hook execution logging - **GitIgnore Pattern**: Fixed gitignore to ensure test-helpers.ts source file is properly tracked ### Changed - **Test Framework**: Upgraded to vitest 3.2.4 with improved performance and compatibility ## [0.2.0] - 2025-08-05 ### Added - **Embedded Hooks System**: Complete TypeScript-based hook system with `claudekit-hooks` CLI - New `claudekit-hooks` binary for running individual hooks with proper TypeScript execution - Hook execution logging and statistics system with `~/.claudekit/logs/` for tracking performance - Support for project-wide and file-specific validation hooks - Built-in hooks: `typecheck-changed`, `check-any-changed`, `lint-changed`, `test-changed`, `create-checkpoint`, `check-todos`, and project-wide variants - **Enhanced Build System**: Separate build targets for main CLI and hooks system with sourcemaps - **ASCII Banner**: Added `oh-my-logo` integration for setup wizard branding - **Hook Commands**: `stats`, `recent`, and `list` subcommands for hook management and monitoring ### Changed - **Hook Configuration**: Simplified matcher syntax in `.claude/settings.json` with pipe separator support (`Write|Edit|MultiEdit`) - **Setup Command**: Enhanced setup wizard with comprehensive hook configuration and visual improvements - **Build Process**: Multi-target build system with dedicated hook compilation and improved TypeScript handling - **Hook Architecture**: All hooks now use embedded TypeScript system instead of external shell scripts ### Removed - **Legacy Shell Hooks**: Removed all shell-based hooks from `src/hooks/` (migrated to TypeScript embedded system) - **Setup Script**: Removed `setup.sh` bash script (functionality moved to TypeScript `setup` command) - **Init Command**: Consolidated `init` functionality into the enhanced `setup` command - **Shell Hook Dependencies**: Eliminated external script dependencies for more reliable hook execution ### Fixed - **Hook Matcher Patterns**: Corrected hook matcher syntax for better tool matching reliability - **Installation Process**: Improved setup wizard with clearer options and error handling - **TypeScript Integration**: Better type safety and error handling across the hook system ### Security - Enhanced allowed-tools validation to prevent unrestricted bash access across all commands ## [0.1.5] - 2025-07-25 ### Documentation - Added command execution guidelines from debugging session to AGENTS.md and create-command.md - Enhanced bash command execution best practices for complex subshells and git commands - Added performance optimization guidelines for combined command execution in slash commands ## [0.1.4] - 2025-07-25 ### Added - Support for aio-stream output formatting in prepare-release script for improved readability when using Claude CLI with stream-json output - Enhanced diff statistics and truncation capabilities in git:commit command - Better test coverage parsing in CI workflow ### Fixed - Corrected allowedTools syntax in prepare-release script (removed unnecessary quotes around tool names) - Updated allowed-tools declarations across multiple commands to use proper bash utility syntax - Resolved failing hook tests by improving test assertions and mock command handling - Fixed test coverage parsing regex pattern in CI workflow - Improved auto-checkpoint test reliability by using file-based logging instead of stderr capture - Fixed command validation test to use proper regex anchoring for frontmatter detection ### Changed - Improved prepare-release script with better error handling and environment validation - Enhanced git:commit command with more detailed diff statistics and smart truncation - Updated test framework to be more robust with better assertion methods