docs: Update CHANGELOG for v0.1.15 release

- Human-in-the-loop approval workflow with Slack reactions
- Conversation memory system for contextual follow-ups
- AgentFlow multi-tenant routing
- Bot self-approval prevention with auto-detected bot_user_id
- Fixed kubectl syntax guidance in agent instructions

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

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

Author: gouravjshah

CHANGELOG.md

@@ -7,30 +7,40 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.1.15] - 2025-12-18
+
 ### Added
-- kubectl-style verb-noun CLI commands for Fleet and Flow resources
-- `Flow` resource type with short name `fw` (alias for workflow execution)
-- `aofctl describe fleet <name>` - Show detailed fleet configuration
-- `aofctl describe flow <name>` - Show detailed flow/workflow configuration
-- `aofctl describe agent <name>` - Show detailed agent configuration
-- Shell completion scaffold (`aofctl completion bash/zsh/fish`)
-- Fleet and Flow command modules (hidden for backwards compatibility)
+- **Human-in-the-Loop Approval Workflow** - Reaction-based command approval for Slack
+  - ✅/❌ reactions for approve/deny destructive operations
+  - Automatic ✅ ❌ reaction buttons added to approval messages
+  - Support for multiple approval reactions (thumbs up, checkmark variants)
+  - Configurable `approval_allowed_users` for role-based approval
+- **Conversation Memory System** - Context persistence across Slack messages
+  - Per-channel and per-thread memory isolation
+  - Automatic context injection for follow-up messages
+  - Supports "delete it", "scale that down" style contextual commands
+  - Configurable memory limits (20 messages default, 10 in context)
+- **AgentFlow Multi-Tenant Routing** - Route messages to different agents based on patterns
+  - Pattern-based agent selection from incoming messages
+  - Support for multiple agents in a single flow
+- **Bot Self-Approval Prevention** - Auto-detects `bot_user_id` at startup
+  - Uses Slack's `auth.test` API to get bot's own user ID
+  - Filters out bot's own reactions to prevent self-approval
+- New documentation guides:
+  - `docs/guides/approval-workflow.md` - Complete approval workflow guide
+  - `docs/guides/conversation-memory.md` - Conversation memory guide
 
 ### Changed
-- **BREAKING**: CLI now uses kubectl-style verb-noun syntax
-  - Old: `aofctl agent run config.yaml`
-  - New: `aofctl run agent config.yaml`
+- kubectl-style verb-noun CLI commands for Fleet and Flow resources
+- `Flow` resource type with short name `fw` (alias for workflow execution)
 - All documentation updated to use `aofctl <verb> <resource>` syntax
-- ADR-001 marked as implemented
-- Migration guide updated with implementation status
+- Updated slack-k8s-bot agent with explicit kubectl syntax guidance
 
 ### Fixed
+- Agent instructions now correctly guide LLM to use `kubectl create deployment` instead of `kubectl run` for deployments with replicas
 - Consistent command syntax across all documentation and examples
 - Example YAML files now use correct `aofctl verb noun` comments
 
-### Deprecated
-- Noun-verb subcommands (`aofctl fleet get`, `aofctl flow run`) - use verb-noun instead
-
 ## [0.1.14] - 2025-12-17
 
 ### Added