🐛 Bug Fixes

Init Command Critical Fix

• Fixed init command configuration templates to generate valid YAML (#279)
  • Corrected field names to match schema (name → title)
  • Fixed configuration structure for branches and tags
  • Added all required fields for proper initialization
  • This fixes a critical issue where new users couldn't properly initialize projects

UI & Accessibility Improvements

• Added proper button type attribute to navigation toggle (#267)
• Fixed theme handling to prevent flash of unstyled content (#266)
• Consolidated duplicate theme management code

📦 Dependencies

• Updated Markdig to 0.41.3
• Updated gitversion.tool to 6.4.0
• Updated Microsoft.Extensions.DependencyInjection to 9.0.8

🙏 Thanks

Thanks to all contributors and users who reported issues!

What's Changed

• fix: improve mobile menu performance and TOC visibility by @pekkah in #265

Full Changelog: 1.1.0...1.2.1

What's Changed

• fix: improve mobile menu performance and TOC visibility by @pekkah in #265

Full Changelog: 1.1.0...1.2.0

What's Changed

• Claude PR Assistant workflow by @pekkah in #256
• Update claude.yml by @pekkah in #259
• feat: implement unified asset handling for xref:// image resolution (Issue #260) by @pekkah in #261
• fix: add mobile hamburger menu and auto-close functionality for left sidebar (fixes #262) by @pekkah in #263

Full Changelog: 1.0.4...1.1.0

What's Changed

Features

• Configurable Link Validation: Added --link-validation parameter to both build and dev commands
  • Build command defaults to strict mode (fails on broken xref references)
  • Dev command defaults to relaxed mode (shows warnings but continues)
  • Broken xrefs in relaxed mode generate placeholder links for better developer experience

Bug Fixes

• Exit Code Propagation: Fixed build failures returning exit code 0 instead of proper error codes (Issue #251)
• Security Enhancement: Improved file filtering for Files sections to prevent unauthorized access (Issue #238)
• XREF Resolution: Enhanced cross-reference link handling for long filenames (Issue #248)

Improvements

• Added consistent error/warning styling throughout the application
• Enhanced development workflow with better error handling in watch mode
• Improved CI/CD compatibility with proper exit codes

Full Changelog: https://github.com/pekka/tanka-docs-gen/compare/1.0.3...1.0.4

🤖 Generated with Claude Code

Co-Authored-By: Claude [email protected]

What's New in v1.0.3

🚀 Critical Fixes

• Fixed build exit codes (#251, #252) - Build failures now properly return non-zero exit codes, allowing CI/CD systems to detect and handle documentation build failures correctly
• Enhanced security filtering (#238, #249) - Implemented comprehensive file filtering to prevent sensitive files (.env, .key, .pem, secrets, etc.) from being exposed in Files sections

🔧 Improvements

• XREF long filename handling (#248, #249) - Verified and added tests to confirm XREF parsing works correctly with long filenames
• CI/CD pipeline enhancements - Improved NuGet publishing and deployment processes

🛡️ Security

• Configurable file security filtering with default patterns for sensitive files
• Priority handling where security filtering overrides hidden file settings
• 45+ comprehensive test cases covering security scenarios

📋 Full Changelog

• Fix build failures returning exit code 0 (#252)
• Fix XREF long filename issue and implement Files section security filtering (#249)
• Restore master branch NuGet publishing for beta versions
• Restrict NuGet publishing to tags only for security

🔄 Migration Notes

• Breaking: None - this is a patch release with backward compatibility
• CI/CD Impact: Build failures will now properly fail CI/CD pipelines (this was the intended behavior)
• Security: Files sections will now automatically exclude sensitive files by default

🏷️ Installation

dotnet tool install --global Tanka.DocsGen --version 1.0.3

Or update existing installation:

dotnet tool update --global Tanka.DocsGen

🤖 Generated with Claude Code

Bug Fixes

Critical Fix: Resolved v1.0.1 regression causing "Cannot access a closed Stream" errors during HTML composition.

What Was Fixed

• Fixed stream disposal issue in DocsMarkdownService where StreamWriter was prematurely closing underlying streams
• Added leaveOpen: true parameter to all StreamReader/StreamWriter constructors
• Preserved UTF-8 encoding support for emoji characters from v1.0.1

Root Cause

The v1.0.1 UTF-8 encoding changes inadvertently altered stream disposal behavior. The StreamWriter(Stream, Encoding) constructor disposes the underlying stream by default, but external components still needed access to those streams.

Testing

• Added comprehensive unit tests reproducing the regression
• Verified fix works with manual documentation builds
• All existing functionality preserved

Impact

This release resolves the critical regression affecting all users who upgraded to v1.0.1. Users experiencing "Cannot access a closed Stream" errors should upgrade immediately.

Fixes: #246

🤖 Generated with Claude Code

Bug Fixes

• Fixed HTML composition failures caused by emoji characters in Markdown files
• Fixed Spectre.Console markup parsing interference with user content
• Added explicit UTF-8 encoding to all file operations
• Enhanced error messages with content preview for debugging

Technical Changes

• Updated GitVersion from 5.8.1 to 6.3.0 with v6 configuration syntax
• Improved error handling in PageComposer and DocsMarkdownService
• Added Markup.Escape() for all user content in console output

🤖 Generated with Claude Code

Full Changelog: 0.9.5...0.10.0