docs: update CHANGELOG.md for v2.1.0 release

Comprehensive changelog covering all changes since v2.0.0:
- 15+ new features (token tracking, Ollama, MCP, skills, etc.)
- 10+ bug fixes (session names, workspace defaults, etc.)
- 5+ major improvements (CLI simplification, init wizard, etc.)
- Breaking changes (DAF_AGENTS.md removal, context simplification)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Author: itdove

CHANGELOG.md

@@ -7,8 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [2.1.0] - 2026-04-02
+
 ### Added
-- **Token usage tracking for AI agent sessions** (#259)
+- **Token usage tracking for AI agent sessions** (#259, #304)
   - Extract token statistics from Claude Code conversation files (.jsonl format)
   - Display token counts in `daf info`, `daf active`, and `daf list` commands
   - Show detailed token breakdown: input tokens, output tokens, cache creation, cache reads
@@ -21,7 +23,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - Token display automatically hidden when agent backend doesn't support tracking
   - `daf list` shows tokens with K/M suffixes for readability (e.g., "1.2M", "850K")
   - Comprehensive cost estimation: `SessionSummary.estimate_cost()` with cache multipliers
-- **Ollama integration for local model support** (#241)
+  - Multi-conversation token aggregation with macOS symlink resolution (#317)
+- **Ollama integration for local model support** (#241, #270, #271)
   - Native `OllamaClaudeAgent` backend using `ollama launch claude` command
   - Zero configuration required - works out of the box with Ollama
   - Automatic server management (no manual server start needed)
@@ -32,19 +35,75 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - Optional `ollama.default_model` config setting (e.g., "qwen3-coder")
   - Simplest local model setup - one install command, no environment variables
   - See [Alternative Model Providers](docs/reference/alternative-model-providers.md) for setup guide
-- **Agent abstraction with `launch_with_prompt()` method**
-  - Added `launch_with_prompt()` to `AgentInterface` for initial prompt support
-  - Implemented in all agents: Claude, Ollama, GitHub Copilot, Cursor, Windsurf
-  - Enables launching agents with initial prompts and session IDs
-  - Skills directories auto-discovery for Claude Code integration
-- **daf-workflow skill** for centralized workflow guidance (#263)
+- **Additional AI agent support** (#274, #302)
+  - Aider agent backend with session management and conversation tracking
+  - Continue agent backend for VS Code integration
+  - Crush AI agent backend with environment variable configuration
+  - All agents support launch, resume, and basic session management
+- **Enterprise model provider enforcement** (#248, #272)
+  - Company-wide enforcement of approved AI model providers
+  - Cost tracking and budget management per profile
+  - Audit logging for compliance and cost attribution
+  - Support for Vertex AI, llama.cpp, and custom providers
+  - TUI integration with enforcement warnings
+  - Model provider profile management commands (#269)
+- **daf-workflow skill** for centralized workflow guidance (#263, #266)
   - Global skill auto-loaded from `~/.claude/skills/daf-workflow/`
   - Comprehensive guidance for AI agents on development workflows
   - Issue tracker auto-detection (GitHub/GitLab/JIRA) based on git remote URLs
   - Guidance on when to use Atlassian MCP vs daf CLI commands (#242)
   - Multi-project session workflows and best practices
   - Standard development and ticket creation session patterns
   - Run `daf upgrade` to install the skill
+- **MCP JIRA integration documentation and skill** (#342)
+  - Comprehensive skill for using Atlassian MCP JIRA tools
+  - Field format requirements and validation guidance
+  - Integration with daf CLI commands
+  - Best practices for MCP vs CLI usage
+- **daf skills command for discovery and inspection** (#340)
+  - List all installed skills with metadata
+  - Show skill content and source locations
+  - Validate skill structure and configuration
+  - Debug skill loading and precedence
+- **Skill system improvements** (#338, #298, #288, #295)
+  - Prevent duplicate skill loading in Claude agent
+  - Multi-agent skill installation with directory mapping
+  - Global and project-level skill installation support
+  - Skills now install to `~/.claude/skills/` by default
+- **Feature orchestration: Multi-session workflow** (#330, #334)
+  - Create coordinated multi-session workflows for complex features
+  - Integrated verification across related sessions
+  - Session dependencies and ordering
+  - Progress tracking and status synchronization
+- **Enhanced CLI parameter support** (#278, #289)
+  - Added CLI parameters for all interactive prompts in session creation
+  - Non-interactive mode support for automation
+  - Full command-line control without TUI prompts
+- **Pre-flight GitHub CLI authentication check** (#277, #284)
+  - Smart failure handling for missing or expired gh authentication
+  - Clear error messages with remediation steps
+  - Prevents cryptic errors during PR/MR creation
+- **Repository selection improvements** (#282)
+  - 3-tier priority system for repository selection
+  - Default prompts with workspace-aware suggestions
+  - Better UX for multi-workspace scenarios
+- **Session goal file support** (#279)
+  - `--goal-file` option to load session goals from files
+  - Supports long-form requirements and specifications
+  - Useful for complex features requiring detailed context
+- **Unified agent management commands** (#251, #275)
+  - Consistent command structure across all agent backends
+  - Unified session management interface
+  - Standardized configuration and status reporting
+- **Agent abstraction with `launch_with_prompt()` method**
+  - Added `launch_with_prompt()` to `AgentInterface` for initial prompt support
+  - Implemented in all agents: Claude, Ollama, GitHub Copilot, Cursor, Windsurf, Aider, Continue, Crush
+  - Enables launching agents with initial prompts and session IDs
+  - Skills directories auto-discovery for Claude Code integration
+- **CLAUDE_CONFIG_DIR environment variable support** (#294)
+  - Override default `~/.claude` directory location
+  - Useful for custom workspace configurations
+  - Supports team-specific Claude Code setups
 
 ### Changed
 - **BREAKING**: Removed DAF_AGENTS.md in favor of daf-workflow skill (#263)
@@ -53,20 +112,92 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - Auto-migration: opening sessions with old DAF_AGENTS.md files prompts for deletion
   - All command prompts updated to reference daf-workflow skill instead of DAF_AGENTS.md
   - Run `daf upgrade` to install daf-workflow skill
+- **Simplified daf init wizard** (#329, #332)
+  - Streamlined onboarding experience
+  - Better user guidance and explanations
+  - Clearer configuration options
+  - Improved validation and error messages
+- **Hierarchical context architecture simplification** (#314, #316)
+  - Use skills-only approach instead of duplicate context files
+  - Eliminates ENTERPRISE.md, ORGANIZATION.md, TEAM.md, USER.md duplication
+  - Skills provide all hierarchical guidance and defaults
+  - Cleaner configuration management
+- **CLI simplification** (#283, #293)
+  - Reduced command count from 73 to ~50 commands
+  - Removed deprecated and redundant commands
+  - Clearer command structure and organization
+  - Better discoverability of available commands
+- **Migrated to pyproject.toml for package configuration** (#261)
+  - Modern Python packaging using PEP 517/518/621
+  - setup.py kept for backward compatibility (no version information)
+  - All package metadata now in pyproject.toml
+  - Cleaner build configuration
+
+### Improved
+- **Branch workflow UX improvements** (#331, #333)
+  - Added missing branch selection prompts
+  - Updated base branch detection logic
+  - Better handling of stale branches
+  - Clearer branch status information
 
 ### Removed
 - DAF_AGENTS.md bundled file - replaced by daf-workflow skill (#263)
   - Session validation now offers to delete old DAF_AGENTS.md files found in repositories
   - Context file loading no longer includes DAF_AGENTS.md
   - All references removed from command prompts and documentation
+- Duplicate context files (ENTERPRISE.md, ORGANIZATION.md, TEAM.md, USER.md) - replaced by skills (#314, #316)
 
 ### Fixed
-- Parent field linking now works without `parent_field_mapping` configuration (#256)
+- **Session names truncated in daf list** (#358, #359)
+  - Fixed truncation making sessions impossible to reopen on small terminals
+  - Improved column width calculations
+  - Better handling of long session names
+- **Auto-convert ticket_creation sessions to development** (#356)
+  - daf sync now automatically converts ticket_creation sessions
+  - Prevents "Session already exists" errors
+  - Smooth workflow transition after ticket creation
+- **Command templates removed from analysis session prompts** (#345)
+  - Analysis sessions no longer show inappropriate code modification commands
+  - Clearer read-only session expectations
+  - Better alignment with investigation/ticket_creation workflows
+- **Branch checkout and sync operation safety checks** (#328)
+  - Added validation before destructive git operations
+  - Prevents accidental branch switches with uncommitted changes
+  - Better error messages for common failure scenarios
+- **Default to session's previous workspace on reopen** (#320, #321)
+  - Session workspace selection now defaults to previous workspace
+  - Improved consistency across session reopens
+  - Better multi-workspace experience
+- **Handle environment variables in complete-on-exit flow** (#318)
+  - Fixed environment variable handling during session completion
+  - Proper cleanup of temporary variables
+  - More robust completion workflow
+- **Resolve macOS symlinks and sum tokens across conversations** (#317)
+  - Fixed token aggregation for sessions with symlinked directories
+  - Proper handling of macOS-specific filesystem features
+  - Accurate token counts in multi-conversation sessions
+- **Skills command naming and type filtering** (#301, #303)
+  - Renamed 'daf skills' to 'daf assets' where appropriate
+  - Added --type parameter for filtering
+  - Clearer command semantics
+- **Remove JIRA-first bias in skills** (#299)
+  - Balanced guidance for all issue tracker types
+  - Equal treatment of GitHub Issues, GitLab Issues, and JIRA
+  - Backend-agnostic workflow documentation
+- **Format parent field as object for API compatibility** (#265)
+  - Fixed JIRA API compatibility for parent field submissions
+  - Proper JSON structure for relationship fields
+  - Better error handling for field format mismatches
+- **Parent field linking without configuration** (#256, #258)
   - `daf jira create --parent` now defaults to standard JIRA `parent` field
   - Previously failed silently when `parent_field_mapping` was not configured
   - Eliminates need to configure `parent_field_mapping` for modern JIRA instances
   - Maintains backward compatibility with legacy configurations using `epic_link`
   - Configuration is now truly optional (only needed for legacy custom fields)
+- **daf complete prompts when no PR should be created** (#252)
+  - Fixed inappropriate PR/MR prompts for ticket_creation and investigation sessions
+  - Session type awareness in completion workflow
+  - Cleaner completion UX
 
 ## [2.0.0] - 2026-03-19
 
@@ -175,6 +306,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - JIRA auto-transition for analysis-only sessions
 - Project path encoding for conversation files
 
-[unreleased]: https://github.com/itdove/devaiflow/compare/v2.0.0...HEAD
+[unreleased]: https://github.com/itdove/devaiflow/compare/v2.1.0...HEAD
+[2.1.0]: https://github.com/itdove/devaiflow/compare/v2.0.0...v2.1.0
 [2.0.0]: https://github.com/itdove/devaiflow/compare/v1.0.0...v2.0.0
 [1.0.0]: https://github.com/itdove/devaiflow/tags/v1.0.0