v1.6.6 - Fix Config Wizard YAML Escaping

v1.6.6 - Fix Config Wizard YAML Escaping

This patch release fixes the remaining YAML parsing error when using the configuration wizard.

Bug Fixed

• Config wizard YAML escaping: Fixed #x001b error when using --wizard or loading wizard-generated configs
  • Root cause: Config wizard was writing ANSI escape codes directly to YAML without proper escaping
  • Symptom: unacceptable character #x001b error when loading .chatrc files created by the wizard
  • Related: This completes the fix started in v1.6.3, which only fixed the default config template

Technical Details

The wizard's _generate_yaml_with_comments() method was writing color values like:

dim: \033[2m    # Wrong - actual ESC character

Now properly escapes them as:

dim: '\\033[2m'  # Correct - will be decoded properly

What Was Fixed

• Wizard now detects ANSI codes containing \033
• Properly escapes backslashes: \ becomes \\\\
• Wraps escaped values in quotes for YAML safety
• YAML parses correctly, decode step converts to ESC character

Testing

• All 318 tests passing
• Clean linting and type checking
• Verified YAML parsing works correctly

Recommendation

If you've experienced #x001b errors when using --wizard or loading configs, upgrade immediately:

pip install --upgrade basic-agent-chat-loop

Then either run --wizard again or --reset-config to regenerate your config file.

v1.6.5 - Flexible Exit Command

v1.6.5 - Flexible Exit Command

This patch release adds flexible exit command handling to prevent common user mistakes.

New Feature

• Flexible exit commands: Now supports multiple variations for exiting the chat
  • exit, quit, bye (no prefix)
  • #exit, #quit, #bye (hash prefix)
  • /exit, /quit, /bye (slash prefix)

Why This Matters

Users often type exit out of habit (like in a terminal), but the CLI previously only recognized #exit. This caused confusion and required retyping with the correct prefix. All variations now work equivalently.

Implementation Details

• Exit command check happens before general command parsing
• Uses lstrip('#/') to normalize input
• Maintains backward compatibility with existing #exit usage

Testing

• All 318 tests passing
• Clean linting and type checking

No upgrade required, but this improves the user experience for anyone who's accidentally typed exit instead of #exit.

v1.6.4 - Critical Performance Fix

v1.6.4 - Critical Performance Fix

This patch release fixes a critical performance bug introduced in v1.6.2 that caused session summaries to be generated after every single turn.

Bug Fixed

• Session summary generation on every turn: Fixed regression where the agent was generating AI summaries after each user query
  • Root cause: After v1.6.2 refactoring, save_conversation() always called _generate_session_summary()
  • Impact: Wasteful API calls, slower response times, confusing summary output in normal conversation
  • Fix: Added generate_summary parameter (default False) to save_conversation()

When Summaries Are Generated

Session summaries are now only generated when explicitly needed:

• ✅ On session compaction (#compact command)
• ✅ On final save when exiting the chat
• ❌ NOT after every query-response turn (fixed!)

Performance Improvements

• Eliminates unnecessary API calls for summary generation
• Faster response times (no summary generation delay after each turn)
• Cleaner conversation flow without spurious summaries

Testing

• All 318 tests passing
• Clean linting and type checking

Recommendation

Users experiencing slow responses or seeing summaries after every turn should upgrade immediately:

pip install --upgrade basic-agent-chat-loop

This release maintains full backward compatibility.

v1.6.3 - Critical Config Fix

v1.6.3 - Critical Config Fix

This patch release fixes a critical bug in v1.6.2 that caused YAML parsing errors when loading configuration files.

Bug Fixed

• YAML parsing error: Fixed "unacceptable character #x001b" error when loading .chatrc files
  • The default config template had improperly escaped ANSI color codes
  • YAML was interpreting escape sequences and creating actual ESC characters (#x001b)
  • Fixed by properly double-escaping backslashes in the config template
  • Config files are now valid YAML while still producing intended ANSI codes

Technical Details

• Changed ANSI codes in default template from \033 to \\033
• This ensures the file contains \033 which YAML parses as literal \033
• The decode('unicode_escape') step then converts to actual ESC characters
• All 331 tests passing

Recommendation

Users experiencing config loading errors should:

1. Upgrade to v1.6.3: pip install --upgrade basic-agent-chat-loop
2. Reset config if needed: chat_loop --reset-config

This release maintains full backward compatibility with properly formatted config files.

v1.6.2 - Code Quality Improvements

v1.6.2 - Code Quality Improvements

This hotfix release includes comprehensive code quality improvements from a systematic QA review:

Code Quality Enhancements

• Eliminated ~380 lines of duplicate code through systematic refactoring
• Created 12 focused helper methods improving code organization and maintainability
• Fixed silent exception handling with proper error logging
• Improved type safety with additional type hints
• Added named constants replacing magic numbers (TOKEN_TO_WORD_RATIO, SECURE_FILE_PERMISSIONS)
• Moved lazy imports to module level for better performance

Refactored Methods

• restore_session(): Reduced from 186 to 80 lines
• _handle_compact_command(): Reduced from 149 to 111 lines
• _extract_token_usage(): Reduced from 104 to 35 lines

Testing

• All 318 tests passing
• Clean linting (ruff)
• Clean type checking (mypy)

This release maintains full backward compatibility while significantly improving code maintainability and debuggability.

v1.6.1 - Incremental Autosave

🔄 Incremental Autosave

Added

• Incremental Autosave - Conversations now save automatically after each message
  • Removed auto_save configuration option (feature is always enabled)
  • Save happens immediately after each query-response cycle
  • Prevents data loss from crashes or force quits
  • All conversations saved to ./.chat-sessions (project-local)
  • Removed manual #save command (no longer needed)

Changed

• Breaking Change: Removed features.auto_save from configuration
  • Auto-save is now always enabled and cannot be disabled
  • Updated help text to reflect always-on autosave
  • Updated configuration examples and documentation

Fixed

• Test suite updated to match removed auto_save configuration
  • All 318 tests passing with new autosave behavior

---

Full Changelog: v1.6.0...v1.6.1

v1.6.0 - Stable Release

🎉 Basic Agent Chat Loop v1.6.0 - Stable Release

This is a stable production release bringing project-local session management and improved UX.

🎯 Breaking Changes

Session Storage Now Project-Local

• Sessions now save to ./.chat-sessions/ in current directory instead of global ~/agent-conversations/
• Provides better context separation between different projects
• Old sessions remain accessible in ~/agent-conversations/ but new sessions use project-local storage
• Configuration option paths.save_location has been removed

Migration Notes: If you have existing sessions in ~/agent-conversations/, they remain accessible but new sessions will be saved to ./.chat-sessions/ in your project directory. To access old sessions, you can either:

1. Manually move them to ./.chat-sessions/ in relevant projects
2. Keep them in the old location for reference (they won't be deleted)

✨ New Features

Improved #resume Command UX

• #resume without arguments now displays available sessions instead of showing an error
• Reduces friction - no need to run #sessions separately first
• Shows helpful usage message after session list

Context Monitoring (from beta.7)

• New #context command shows token usage and session statistics
• Status bar shows context usage percentage
• Configurable warning thresholds via context.warning_thresholds in config
• Auto-suggest #compact command when approaching context limits

🛠️ Improvements

Config Wizard Updates

• Updated help text to reference new ./.chat-sessions location
• Removed save_location prompts from wizard flow
• All documentation updated to reflect project-local session storage

Token Tracking (from beta.7)

• Fixed AWS Strands token counting accuracy
• Proper handling of streaming response token counts
• Better aggregate token statistics

📝 What's Changed Since v1.5.1

v1.6.0-beta.8 (2025-12-30)

• fix: update config wizard help text to reference ./.chat-sessions
• feat: change session storage to project-local ./.chat-sessions
• feat: improve #resume command UX to show sessions when no arg provided

v1.6.0-beta.7 (2025-12-30)

• feat: add context usage monitoring with #context command
• feat: add configurable context warning thresholds
• feat: display context % in status bar
• fix: accurate token counting for AWS Strands agents

📦 Installation

pip install basic-agent-chat-loop==1.6.0

Or upgrade from previous version:

pip install --upgrade basic-agent-chat-loop

📚 Documentation

• README: View on GitHub
• Configuration Guide: docs/CONFIG.md
• Changelog: CHANGELOG.md

🙏 Thank You

Thank you to all users who tested the beta releases and provided feedback!

---

Full Changelog: v1.5.1...v1.6.0

v1.6.0-beta.8

What's New in v1.6.0-beta.8

🎯 Breaking Changes

Session Storage Now Project-Local (#BREAKING)

• Sessions now save to ./.chat-sessions/ in current directory instead of global ~/agent-conversations/
• Provides better context separation between different projects
• Old sessions remain accessible in ~/agent-conversations/ but new sessions use project-local storage
• Configuration option paths.save_location has been removed

✨ Features

Improved #resume Command UX (#enhancement)

• #resume without arguments now displays available sessions instead of showing an error
• Reduces friction - no need to run #sessions separately first
• Shows helpful usage message after session list

🛠️ Fixes

Config Wizard Updates (#fix)

• Updated help text to reference new ./.chat-sessions location
• Removed save_location prompts from wizard flow
• All documentation updated to reflect project-local session storage

📝 Changes Since v1.6.0-beta.7

Full commit history:

• c1d7a02: fix: update config wizard help text to reference ./.chat-sessions
• 058f6ab: feat: change session storage to project-local ./.chat-sessions
• 6a29eab: feat: improve #resume command UX to show sessions when no arg provided

📦 Installation

pip install basic-agent-chat-loop==1.6.0b8

⚠️ Migration Notes

If you have existing sessions in ~/agent-conversations/, they will remain accessible but new sessions will be saved to ./.chat-sessions/ in your project directory. To access old sessions, you can either:

1. Manually move them to ./.chat-sessions/ in relevant projects
2. Keep them in the old location for reference (they won't be deleted)

The new project-local approach provides better separation between different project contexts.

v1.6.0-beta.7 - Context Monitoring & Token Tracking Fix

🎯 Context Monitoring Features

Added

• #context Command - Display detailed token usage statistics

  • Shows total, input, and output tokens with formatted display (K/M suffixes)
  • Shows percentage of context used when max_tokens is known
  • Displays session time and query count
  • Includes configurable warning messages at threshold percentages

• Status Bar Context Percentage - Real-time context usage display

  • Example: 45.2K tokens (23%) or 180.5K tokens (90%)
  • Dynamically formats large numbers (K for thousands, M for millions)

• Configurable Warning Thresholds

  • Configure via context.warning_thresholds in .chatrc
  • Default thresholds: [80, 90, 95]
  • Highest threshold (95%) shows error with #compact suggestion
  • Second-highest (90%) shows warning message
  • Lower thresholds (80%) show informational messages
  • Supports per-agent customization

Configuration Example

```yaml

Default configuration

context:
warning_thresholds: [80, 90, 95]

Custom per-agent

agents:
'Memory Heavy Agent':
context:
warning_thresholds: [60, 80, 90, 95] # Warn earlier
```

🐛 Bug Fixes

AWS Strands Token Accumulation Fix

• Fixed critical bug where token counts could exceed context limits
• AWS Strands accumulated_usage is cumulative across entire session
• Implemented delta calculation to track only per-query token usage
• Added (usage_dict, is_accumulated) tuple return from _extract_token_usage()
• Prevents impossible token counts (e.g., user-reported "70.6M tokens" in status bar)

📚 Documentation

• Updated CHANGELOG.md with new features and bug fix
• Added #context to README.md commands table
• Added comprehensive context section to docs/CONFIG.md

🔗 Related Issues

• Closes #48 - Conversation compaction to manage context limits (core monitoring features)

📦 Installation

```bash
pip install --upgrade basic-agent-chat-loop==1.6.0b7
```

🧪 Testing

• ✅ All 318 tests passing
• ✅ Ruff linting passed
• ✅ Mypy type checking passed

v1.6.0-beta.6

What's Changed

Changed

• Command Prefix Update - All in-chat commands now require # prefix to avoid collisions with regular conversation
  • Commands updated: #help, #info, #compact, #sessions, #save, #copy, #resume, #clear, #quit, #exit
  • Template commands still use / prefix (intentional)
  • Multi-line input still uses \\ (intentional)
  • Added helpful error message for unknown # commands
  • Updated all help text and README documentation

Fixed

• Exit Command Hanging - Added explicit sys.exit(0) to ensure clean terminal return
  • Fixed issue where #exit would show "[Process Completed]" but not return to prompt
  • Process now terminates cleanly and returns control immediately

Installation

pip install basic-agent-chat-loop==1.6.0b6

Full Changelog: v1.6.0-beta.5...v1.6.0-beta.6