Release v0.1.3

Changes in v0.1.3

Fixed

• Script permissions for integration tests
• Repository URLs updated to d-oit organization
• Common logging functionality improvements
• ShellCheck warnings in CI (exported LOG_FILE and ERROR_LOG_FILE)
• GitHub release workflow paths (plugin.json → .claude-plugin/plugin.json)
• GitHub release workflow to use CHANGELOG.md instead of CHANGES.md

Changed

• Updated owner/contributor information to d-oit organization
• Tests now self-contained with automatic cleanup

Improved

• Integration test reliability (test_cache_stats, test_analytics_report)
• Refactored common logging into shared module (scripts/common-logging.sh)
• All shell scripts now pass ShellCheck validation in CI

---

Installation:

cd ~/.config/claude-code/plugins
git clone https://github.com/d-oit/gemini-search-plugin.git gemini-search

Usage:

• /search [query] - Perform web searches
• /search-stats - View usage statistics
• /clear-cache - Clear search cache

Release v0.1.2 - Skills System & Code Quality

Skills System & Code Quality Improvements

Added

• Skills System: Four comprehensive Claude Code skills for plugin development workflow
  • plugin-creator: Complete guide for creating Claude Code plugins with structure, commands, agents, hooks
  • shell-script-quality: ShellCheck and BATS testing framework integration and best practices
  • github-repo-management: GitHub repository management including issues, PRs, releases, and CI/CD
  • web-research: Token-efficient web research using gemini-search agent (30-40% token savings)
• Skills documentation with README.md files for each skill
• Main .claude/skills/README.md with comprehensive overview and integration guide
• Cross-references between skills for integrated workflow

Fixed

• Fixed ShellCheck warnings in hooks/error-search.sh (SC2221/SC2222 pattern matching)
• Fixed ShellCheck info in scripts/search-wrapper.sh (SC1091 with proper directive)
• Fixed ShellCheck warning in tests/run-integration-tests.sh (SC2155 declare and assign separately)
• Fixed ShellCheck info in tests/test-link-validation.sh (SC1091 with proper directive)

Improved

• All shell scripts now pass ShellCheck validation without warnings
• Enhanced code quality with proper shellcheck directives
• Better test script quality assurance
• Comprehensive skills-based workflow for plugin development

Documentation

• Added skill-specific README files for quick reference
• Updated plugin-creator skill with related skills references
• Complete web-research skill documentation with usage patterns and examples
• GitHub repository management skill with CI/CD workflow examples
• Shell script quality skill with complete testing and linting guide

Installation

```bash
/plugin add https://github.com/d-oit/gemini-search-plugin
```

Full Changelog

See CHANGELOG.md for complete version history.

Release 0.1.0

Initial Release

This is the first public release of the Gemini Search Plugin for Claude Code.

Added

Core Search Functionality

• Integrated Gemini CLI in headless mode (gemini -p with --yolo flag)
• Restricted Gemini CLI to only use google_web_search tool via .gemini/settings.json
• Zero direct web scraping or HTTP requests - all content via Gemini's grounded search

Caching System

• Smart caching with 1-hour TTL
• MD5-based cache keys for efficient lookups
• Platform-independent cache validation (supports macOS, Linux, Windows)
• Cache statistics and management commands

Analytics & Tracking

• Comprehensive analytics tracking with JSON storage
• Cache hit/miss rate monitoring
• Search engine usage distribution
• Top 10 queries tracking (automatically maintained)
• Token savings estimates (39% per cache hit)
• Atomic writes to prevent data corruption

Content Extraction

• Dynamic content extraction from web pages using Gemini CLI
• Clean text extraction (removes HTML, scripts, styling)
• Content relevance validation
• Multiple content sources with fallback methods

Validation Pipeline

• Multi-stage result validation to prevent false positives
• Query term matching with relevance scoring
• Domain blacklist filtering (example.com, test.com, etc.)
• Content quality checks (minimum length, error indicators)
• 50% minimum relevance threshold for search results
• 30% minimum relevance threshold for extracted content

Commands

• /search [query] - Perform web search with caching and validation
• /search-stats - View analytics and cache statistics
• /clear-cache - Clear cache and reset analytics

Hooks

• error-search.sh - Auto-detects errors with exponential backoff retry logic
• pre-edit-search.sh - Provides search suggestions before edits

Agents

• gemini-search - Isolated context subagent for search operations (39% token savings)

Error Handling

• Exponential backoff retry logic (configurable via environment variables)
• Error type categorization (timeout, not_found, server_error, rate_limit, etc.)
• Comprehensive error logging with structured JSON format
• Graceful degradation when services unavailable

Development & Testing

• Integration test suite with 10 automated tests
• ShellCheck linting for all bash scripts
• JSON validation for all configuration files
• CI/CD workflows (GitHub Actions)
• Release automation with version validation

Security Features

• Content size limits (100KB default, configurable)
• URL validation to prevent injection
• Timeout protection (15s default)
• Error message sanitization
• No hardcoded credentials
• Comprehensive SECURITY.md documentation

Documentation

• Complete README.md with features and usage
• CONTRIBUTING.md with development guidelines
• SECURITY.md with vulnerability reporting
• DEPLOYMENT.md with testing and release procedures
• Inline documentation in all scripts

Repository Setup

• Branch protection rules on master branch
• Issue and PR templates
• Automated CI/CD pipeline
• Security scanning (Trivy)
• Semantic versioning support

Configuration

Environment variables for customization:

• CACHE_DIR - Cache directory location
• CACHE_TTL - Cache time-to-live in seconds (default: 3600)
• LOG_FILE - Main log file path
• ERROR_LOG_FILE - Error log path
• MAX_RETRIES - Retry attempts on failure (default: 3)
• RETRY_DELAY - Initial retry delay in seconds (default: 1)
• BACKOFF_BASE - Exponential backoff multiplier (default: 2)
• EXPONENTIAL_MAX_DELAY - Maximum retry delay cap (default: 60)
• ANALYTICS_DIR - Analytics data directory
• TIMEOUT_SECONDS - Content extraction timeout (default: 15)
• MAX_CONTENT_SIZE - Maximum content size to process (default: 100000)

Dependencies

• Required: Gemini CLI (npm install -g @google/genai-cli) - must be pre-authenticated
• Required: jq for JSON processing
• Required: md5sum for cache key generation
• Required: stat for cache TTL validation

Note: The plugin uses the Gemini CLI's existing authentication. No separate API key configuration is needed for the plugin itself.

Known Limitations

• Cache stored in /tmp by default (consider private location on shared systems)
• Log files may contain sensitive query information
• Requires Gemini CLI to be pre-authenticated (gemini auth login or equivalent)

---

Release Checklist

For maintainers preparing releases:

• Update version in .claude-plugin/plugin.json
• Update version in .claude-plugin/marketplace.json
• Update CHANGES.md with release date
• Run integration tests: bash tests/run-integration-tests.sh
• Commit changes: git commit -m "chore: prepare release vX.Y.Z"
• Create tag: git tag -a vX.Y.Z -m "Release vX.Y.Z"
• Push: git push origin master && git push origin vX.Y.Z

---

Full Changelog: https://github.com/d-oit/gemini-search-plugin/commits/v0.1.0