sqlew 2.1.3 → 3.0.2

package/README.md

@@ -1,613 +1,302 @@
-# sqlew
-![sqlew_logo](assets/sqlew-logo.png)
-
-[![npm version](https://img.shields.io/npm/v/sqlew.svg)](https://www.npmjs.com/package/sqlew)
-[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
-
-> **SQL Efficient Workflow** - MCP server for efficient context sharing between Claude Code sub-agents
-
-## Overview
-
-**sqlew** is a Model Context Protocol (MCP) server that enables efficient context sharing between multiple Claude Code agents through a SQLite-backed database. It dramatically reduces token consumption while providing structured data management through metadata-driven organization.
-
-**Current Version:** 2.1.1
-
-## Benefits
-
-### 🧠 Organizational Memory for AI Agents
-
-**sqlew solves the "organizational memory" problem that traditional code can't:**
-
-| What Traditional Code Provides | What sqlew Adds |
-|-------------------------------|-----------------|
-| ✅ **Git history** - WHAT changed | ✅ **Decisions** - WHY it changed |
-| ✅ **Code comments** - HOW it works | ✅ **Constraints** - WHY it must work this way |
-| ❌ **Architectural decisions** - Missing! | ✅ **Context survival** - Across sessions |
-
-**Real-World Example:**
-```typescript
-// Agent in Session 1 records:
-{
-  key: "loom/duration-constraint",
-  value: "Duration must NOT occur in Loom module",
-  layer: "business",
-  tags: ["architecture", "constraint", "breaking"]
-}
-
-// Agent in Session 2 queries:
-"What are business layer constraints for Loom module?"
-→ Finds: "Duration must NOT occur in Loom"
-→ Avoids introducing the same bug you just fixed!
-```
-
-### 💡 Why This Matters for LLMs
-
-1. **Context Survival**: Next Claude session can query architectural decisions from previous sessions
-2. **Prevents Regression**: Constraints like "Duration must NOT occur in Loom" prevent reintroducing bugs
-3. **Historical Reasoning**: Captures WHY decisions were made, not just WHAT changed
-4. **Knowledge Discovery**: Searchable by layer/tag/scope - "Show me all breaking changes in business layer"
-
-### ⚡ Real-World Token Savings
-
-**Scenario: 5 Agents Working Across 10 Sessions**
-
-| Approach | Token Usage | Details |
-|----------|-------------|---------|
-| **Without sqlew** | ~20,000 tokens | All context re-provided every session |
-| **With sqlew** | ~7,400 tokens | Selective queries, persistent storage |
-| **Savings** | **63% reduction** | Realistic multi-session project |
-
-**Why sqlew Saves Tokens:**
-
-1. **Selective Retrieval** (50-70% savings)
-   - Without: Must read ALL context every time
-   - With: Query only what's needed - `search_layer("business")` returns 5 decisions, not 100
-
-2. **Structured vs Unstructured** (60-85% savings)
-   - Without: "We decided to use JWT because..." (50-200 tokens in prose)
-   - With: `{key: "auth", value: "JWT", layer: "business"}` (20-30 tokens)
-
-3. **Cross-Session Persistence** (80-95% savings)
-   - Without: Context re-provided every new session
-   - With: Query from database, context survives sessions
-
-4. **Search Instead of Scan** (70-90% savings)
-   - Without: Read all 100 decisions to find 5 relevant ones
-   - With: `search_tags(["breaking", "api"])` → only relevant results
-
-**Expected Token Reduction (Typical Multi-Agent Project):**
-- Conservative: 50-65% reduction
-- Realistic: 60-75% reduction
-- Optimal: 70-85% reduction
-
-*Note: Internal architecture improvements (v1.0.0→v2.0.0) achieved 96% tool definition reduction and 67% MCP context reduction. The percentages above reflect real-world usage benefits.*
-
-**Performance:**
-- Query response: 2-50ms
-- Concurrent access: 5+ simultaneous agents
-- Storage efficiency: ~140 bytes per decision
-
-**Reliability:**
-- SQLite ACID transactions
-- 100% backward compatible upgrades
-
-## Features
-
-### Core Capabilities
-- **Context Management** - Store and retrieve decisions with tags, layers, scopes, and versions
-- **Agent Messaging** - Priority-based messaging system with broadcast support
-- **File Change Tracking** - Layer-based file organization with lock detection
-- **Constraint Management** - Track performance, architecture, and security constraints
-- **Activity Logging** - Automatic tracking of all agent actions (v2.1.0)
-- **Weekend-Aware Auto-Cleanup** - Smart retention policies that pause during weekends
-
-### Advanced Features (v2.1.0)
-- **Smart Defaults** - Auto-infer layer and tags from file paths (60% token reduction)
-- **Batch Operations** - Process up to 50 operations atomically (70% token reduction)
-- **Update Polling** - Lightweight subscription mechanism (95% token reduction)
-- **Advanced Query** - Complex multi-criteria filtering with full-text search
-- **Templates** - 5 built-in templates + custom template support
-- **CLI Tool** - Standalone query commands without MCP server
-
-### 6 Consolidated Tools (26 Actions)
-1. **`decision`** (13 actions) - Context management with metadata
-2. **`message`** (4 actions) - Agent-to-agent messaging
-3. **`file`** (4 actions) - File change tracking
-4. **`constraint`** (3 actions) - Constraint management
-5. **`stats`** (4 actions) - Statistics and cleanup
-6. **`config`** (2 actions) - Configuration management
-
-## Installation
-
-### Requirements
-- Node.js 18.0.0 or higher
-- npm or npx
-
-### Install from npm
-
-```bash
-npm install sqlew
-```
-
-### Quick Test
-
-```bash
-# Test with MCP Inspector
-npx @modelcontextprotocol/inspector npx sqlew
-```
-
-## Quick Start
-
-### Add to Claude Desktop
-
-Add to your Claude Desktop configuration (`claude_desktop_config.json`):
-
-```json
-{
-  "mcpServers": {
-    "sqlew": {
-      "command": "npx",
-      "args": ["sqlew"]
-    }
-  }
-}
-```
-
-### Custom Database Path
-
-```json
-{
-  "mcpServers": {
-    "sqlew": {
-      "command": "npx",
-      "args": ["sqlew", "/path/to/database.db"]
-    }
-  }
-}
-```
-
-### Weekend-Aware Auto-Deletion
-
-```json
-{
-  "mcpServers": {
-    "sqlew": {
-      "command": "npx",
-      "args": [
-        "sqlew",
-        "--autodelete-ignore-weekend",
-        "--autodelete-message-hours=48",
-        "--autodelete-file-history-days=10"
-      ]
-    }
-  }
-}
-```
-
-### Default Database Location
-
-`.sqlew/sqlew.db` (created in current directory)
-
-## MCP Tools Reference
-
-All tools use action-based routing. Call any tool with `action: "help"` for comprehensive documentation.
-
-### 1. `decision` - Context Management
-
-Manage decisions with metadata (tags, layers, versions, scopes).
-
-**Actions:** `set`, `get`, `list`, `search_tags`, `search_layer`, `versions`, `quick_set`, `search_advanced`, `set_batch`, `has_updates`, `set_from_template`, `create_template`, `list_templates`, `help`
-
-**Examples:**
-
-```typescript
-// Standard set
-{
-  action: "set",
-  key: "auth_method",
-  value: "JWT",
-  agent: "auth-agent",
-  layer: "business",
-  tags: ["authentication", "security"],
-  version: "1.0.0"
-}
-
-// Quick set with smart defaults (auto-infers layer, tags)
-{
-  action: "quick_set",
-  key: "api/users/auth",
-  value: "JWT validation updated",
-  agent: "auth-agent"
-  // Auto-infers: layer="presentation", tags=["api", "users", "auth"]
-}
-
-// Batch set (up to 50 decisions atomically)
-{
-  action: "set_batch",
-  decisions: [
-    { key: "api_v1", value: "REST", layer: "presentation" },
-    { key: "api_v2", value: "GraphQL", layer: "presentation" }
-  ],
-  atomic: true
-}
-
-// Check for updates (lightweight polling)
-{
-  action: "has_updates",
-  agent_name: "my-agent",
-  since_timestamp: "2025-10-15T10:00:00Z"
-}
-
-// Advanced search with complex filtering
-{
-  action: "search_advanced",
-  layers: ["business", "data"],
-  tags_all: ["security"],
-  search_text: "authentication",
-  status: "active",
-  limit: 20
-}
-
-// Set from template
-{
-  action: "set_from_template",
-  template: "api-endpoint",
-  key: "user_api",
-  value: "GET /api/users"
-}
-
-// Get decision
-{
-  action: "get",
-  key: "auth_method"
-}
-
-// Get version history
-{
-  action: "versions",
-  key: "auth_method"
-}
-
-// Get help
-{ action: "help" }
-```
-
-### 2. `message` - Agent Messaging
-
-Send and retrieve messages with priority levels.
-
-**Actions:** `send`, `get`, `mark_read`, `send_batch`, `help`
-
-**Examples:**
-
-```typescript
-// Send message
-{
-  action: "send",
-  from_agent: "agent1",
-  to_agent: "agent2",
-  msg_type: "warning",
-  message: "File locked",
-  priority: "high",
-  payload: { file: "/src/auth.ts" }
-}
-
-// Batch send (up to 50 messages)
-{
-  action: "send_batch",
-  messages: [
-    {
-      from_agent: "orchestrator",
-      to_agent: "auth-agent",
-      msg_type: "request",
-      message: "Start setup",
-      priority: "high"
-    },
-    {
-      from_agent: "orchestrator",
-      to_agent: "db-agent",
-      msg_type: "request",
-      message: "Initialize schema",
-      priority: "high"
-    }
-  ]
-}
-
-// Get messages
-{
-  action: "get",
-  agent_name: "agent1",
-  unread_only: true,
-  priority_filter: "high"
-}
-
-// Mark as read
-{
-  action: "mark_read",
-  agent_name: "agent1",
-  message_ids: [1, 2, 3]
-}
-```
-
-### 3. `file` - File Change Tracking
-
-Track file modifications with layer assignment.
-
-**Actions:** `record`, `get`, `check_lock`, `record_batch`, `help`
-
-**Examples:**
-
-```typescript
-// Record file change
-{
-  action: "record",
-  file_path: "/src/auth.ts",
-  agent_name: "auth-agent",
-  change_type: "modified",
-  layer: "business",
-  description: "Updated JWT validation"
-}
-
-// Batch record (up to 50 file changes)
-{
-  action: "record_batch",
-  file_changes: [
-    {
-      file_path: "/src/auth/jwt.ts",
-      agent_name: "auth-agent",
-      change_type: "created",
-      layer: "business"
-    },
-    {
-      file_path: "/src/auth/validation.ts",
-      agent_name: "auth-agent",
-      change_type: "created",
-      layer: "business"
-    }
-  ]
-}
-
-// Get file changes
-{
-  action: "get",
-  since: "2025-10-15T10:00:00Z",
-  layer: "business"
-}
-
-// Check file lock
-{
-  action: "check_lock",
-  file_path: "/src/auth.ts",
-  lock_duration: 300
-}
-```
-
-### 4. `constraint` - Constraint Management
-
-Manage architectural, performance, and security constraints.
-
-**Actions:** `add`, `get`, `deactivate`, `help`
-
-**Examples:**
-
-```typescript
-// Add constraint
-{
-  action: "add",
-  category: "performance",
-  constraint_text: "Response time < 200ms",
-  priority: "high",
-  layer: "business",
-  tags: ["api", "performance"]
-}
-
-// Get constraints
-{
-  action: "get",
-  category: "performance",
-  active_only: true
-}
-
-// Deactivate constraint
-{
-  action: "deactivate",
-  constraint_id: 42
-}
-```
-
-### 5. `stats` - Statistics & Utilities
-
-Get database statistics and manage data cleanup.
-
-**Actions:** `layer_summary`, `db_stats`, `clear`, `activity_log`, `help`
-
-**Examples:**
-
-```typescript
-// Layer summary
-{ action: "layer_summary" }
-
-// Database stats
-{ action: "db_stats" }
-
-// Activity log (v2.1.0)
-{
-  action: "activity_log",
-  since: "1h",
-  agent_names: ["auth-agent"],
-  limit: 100
-}
-
-// Clear old data (weekend-aware)
-{
-  action: "clear",
-  messages_older_than_hours: 48,
-  file_changes_older_than_days: 14
-}
-```
-
-### 6. `config` - Configuration
-
-Manage auto-deletion and retention settings.
-
-**Actions:** `get`, `update`, `help`
-
-**Examples:**
-
-```typescript
-// Get config
-{ action: "get" }
-
-// Update config
-{
-  action: "update",
-  ignoreWeekend: true,
-  messageRetentionHours: 48,
-  fileHistoryRetentionDays: 10
-}
-```
-
-## CLI Tool (v2.1.0)
-
-Query your database directly from terminal without MCP server.
-
-### Available Commands
-
-```bash
-# Query decisions
-npx sqlew-cli query decisions --layer=business --tags=breaking --output=table
-
-# Query messages
-npx sqlew-cli query messages --unread --priority=high --output=json
-
-# Query file changes
-npx sqlew-cli query files --since=1h --layer=data --output=table
-
-# Query activity log
-npx sqlew-cli query activity --since=5m --agent=* --output=json
-```
-
-### Common Options
-
-- `--output <format>` - Output format: `json` or `table` (default: json)
-- `--layer <layer>` - Filter by layer
-- `--tags <tags>` - Filter by tags (comma-separated)
-- `--since <time>` - Time filter (e.g., "5m", "1h", "2d", or ISO timestamp)
-- `--limit <number>` - Limit results
-- `--db-path <path>` - Custom database path
-- `--help` - Show help
-
-## Architecture Layers
-
-sqlew organizes code by standard architecture layers:
-
-- **presentation** - UI, API endpoints, views
-- **business** - Business logic, services, use cases
-- **data** - Repositories, database access
-- **infrastructure** - Configuration, external services
-- **cross-cutting** - Logging, security, utilities
-
-## Database Schema
-
-**Master Tables (m_ prefix):** Normalization layer (agents, files, keys, layers, tags, scopes, categories, config, templates)
-
-**Transaction Tables (t_ prefix):** Core data (decisions, history, messages, file changes, constraints, activity log)
-
-**Views (v_ prefix):** Token-efficient pre-aggregated queries
-
-**Triggers (trg_ prefix):** Automatic version history and activity logging
-
-### Automatic Migration
-
-- **v1.x → v2.x:** Automatic migration adds table prefixes and new features
-- **v2.0 → v2.1:** Adds activity log and template tables
-- All migrations are safe with rollback protection
-
-## Development
-
-### Building from Source
-
-```bash
-git clone https://github.com/sin5ddd/mcp-sqlew.git
-cd mcp-sqlew
-npm install
-npm run build
-```
-
-### Available Scripts
-
-```bash
-npm start          # Start MCP server
-npm run cli        # Run CLI tool
-npm run inspector  # Test with MCP Inspector
-npm run build      # Build TypeScript
-npm run dev        # Watch mode
-npm run rebuild    # Clean and rebuild
-```
-
-## Configuration
-
-### Retention Periods (Defaults)
-
-- **Messages:** 24 hours (weekend-aware optional)
-- **File Changes:** 7 days (weekend-aware optional)
-- **Decisions:** Permanent (version history preserved)
-- **Constraints:** Permanent (soft delete only)
-
-### Weekend-Aware Cleanup
-
-When enabled, weekends (Saturday/Sunday) don't count toward retention periods:
-
-- Message sent Friday 3pm → Deleted Monday 3pm (skips weekend)
-- Message sent Monday 10am → Deleted Tuesday 10am
-
-Configure via CLI args or MCP tools at runtime.
-
-## Migration Guide
-
-### From v2.1.0 to v2.1.1
-
-No breaking changes. Only bin command configuration changed:
-
-- **Old:** `npx sqlew` → CLI, `npx sqlew-server` → MCP server
-- **New:** `npx sqlew` → MCP server (default), `sqlew-cli` → CLI (after installing the package)
-
-Update Claude Desktop config if using custom commands.
-
-### From v2.0.0 to v2.1.0
-
-No breaking changes. Database migrates automatically on startup.
-
-New features are opt-in via new actions.
-
-### From v1.x to v2.0.0
-
-Requires migration. See [MIGRATION_v2.md](MIGRATION_v2.md) for details.
-
-## License
-
-MIT - see [LICENSE](LICENSE) file for details
-
-## Links
-
-- [npm package](https://www.npmjs.com/package/sqlew)
-- [GitHub repository](https://github.com/sin5ddd/mcp-sqlew)
-- [Changelog](CHANGELOG.md)
-- [Architecture Documentation](ARCHITECTURE.md)
-- [Model Context Protocol](https://modelcontextprotocol.io/)
-
-## Author
-
-**sin5ddd**
-
-## Support
-
-- **Issues:** [GitHub Issues](https://github.com/sin5ddd/mcp-sqlew/issues)
-- **Documentation:** See [ARCHITECTURE.md](ARCHITECTURE.md) for technical details
-
-## Acknowledgments
-
-Built with:
-- [Model Context Protocol SDK](https://github.com/modelcontextprotocol/sdk)
-- [better-sqlite3](https://github.com/WiseLibs/better-sqlite3)
-- TypeScript
+# sqlew
+![sqlew_logo](assets/sqlew-logo.png)
+
+[![npm version](https://img.shields.io/npm/v/sqlew.svg)](https://www.npmjs.com/package/sqlew)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
+> **SQL Efficient Workflow** - MCP server for efficient context sharing between Claude Code sub-agents
+
+## What is sqlew?
+
+**sqlew** is a Model Context Protocol (MCP) server that gives AI agents organizational memory. It solves a critical problem: **AI agents can't remember decisions across sessions**.
+
+### The Problem
+
+Without sqlew:
+- 🔴 Every new Claude session starts with zero context
+- 🔴 Must re-explain architectural decisions every time
+- 🔴 Agents can reintroduce bugs that were already fixed
+- 🔴 No way to track WHY decisions were made
+
+### The Solution
+
+With sqlew, AI agents can:
+- ✅ **Remember** architectural decisions across sessions
+- ✅ **Query** past context: "What breaking changes were made to the API?"
+- ✅ **Prevent** regressions by storing constraints: "Duration must NOT occur in Loom module"
+- ✅ **Coordinate** between multiple agents with messaging and task tracking
+
+**Real-World Example:**
+```typescript
+// Agent in Session 1 records:
+{
+  key: "api_v1_deprecated",
+  value: "/v1/users endpoint deprecated, use /v2/users",
+  tags: ["api", "breaking-change"]
+}
+
+// Agent in Session 2 (days later) queries:
+"What API changes should I know about?"
+→ Finds the deprecation decision
+→ Uses /v2/users endpoint automatically!
+```
+
+## Why Use sqlew?
+
+### 🧠 Organizational Memory
+Traditional code only tells you **WHAT** and **HOW**. sqlew adds **WHY**:
+- Git history → WHAT changed
+- Code comments → HOW it works
+- **sqlew decisions** → **WHY** it was changed
+- **sqlew constraints** → **WHY** it must work this way
+
+### ⚡ Token Efficiency
+**60-75% token reduction** in multi-session projects:
+- Store decisions once, query selectively
+- Structured data (20-30 tokens) vs prose (50-200 tokens)
+- Cross-session persistence eliminates context re-explanation
+
+### 🎯 Key Features
+- **7 Specialized Tools**: decisions, messages, tasks, files, constraints, stats, config
+- **Metadata-Driven**: Tag, layer, scope, and version everything
+- **Auto-Stale Detection**: Tasks automatically transition when idle
+- **Weekend-Aware Cleanup**: Smart retention that pauses during weekends
+- **Batch Operations**: Process up to 50 items atomically
+
+### 🔖Kanban-style AI Scrum
+![kanban-style task management](assets/kanban-style.png)
+
+## Installation
+
+### Requirements
+- Node.js 18.0.0 or higher
+- npm or npx
+
+### Quick Install
+
+```bash
+npm install sqlew
+```
+
+### Add to Claude Desktop
+
+Edit `claude_desktop_config.json`:
+
+```json
+{
+  "mcpServers": {
+    "sqlew": {
+      "command": "npx",
+      "args": ["sqlew"]
+    }
+  }
+}
+```
+
+That's it! Restart Claude Desktop and sqlew is ready.
+
+### Custom Database Path (Optional)
+
+```json
+{
+  "mcpServers": {
+    "sqlew": {
+      "command": "npx",
+      "args": ["sqlew", "/path/to/database.db"]
+    }
+  }
+}
+```
+
+Default location: `.sqlew/sqlew.db` in current directory
+
+## Quick Start
+
+### For AI Agents
+
+⚠️ **Most Important Rule**: ALWAYS include the `action` parameter in every tool call.
+
+```javascript
+// ❌ WRONG
+{key: "auth_method", value: "jwt"}
+
+// ✅ CORRECT
+{action: "set", key: "auth_method", value: "jwt", layer: "business"}
+```
+
+### Basic Usage
+
+```javascript
+// Store a decision
+{
+  action: "set",
+  key: "auth_method",
+  value: "JWT with refresh tokens",
+  layer: "business",
+  tags: ["authentication", "security"]
+}
+
+// Query it later
+{
+  action: "get",
+  key: "auth_method"
+}
+
+// Search by tags
+{
+  action: "search_tags",
+  tags: ["security"],
+  status: "active"
+}
+```
+
+### Available Tools
+
+| Tool | Purpose | Example Use |
+|------|---------|------------|
+| **decision** | Record choices | "We chose PostgreSQL" |
+| **message** | Agent communication | "Task completed" |
+| **task** | Track work | "Implement feature X" |
+| **file** | Track changes | "Modified auth.ts" |
+| **constraint** | Define rules | "API must be <100ms" |
+| **stats** | Database metrics | Get layer summary |
+| **config** | Retention settings | Auto-cleanup config |
+
+Each tool supports `action: "help"` for full documentation and `action: "example"` for comprehensive usage examples.
+
+## Documentation
+
+### On-Demand Documentation
+
+**All tools provide built-in documentation with zero upfront token cost:**
+- `action: "help"` - Detailed parameter reference, action descriptions, examples
+- `action: "example"` - Comprehensive usage scenarios, workflows, best practices
+
+**Example:**
+```javascript
+// Get detailed help for decision tool
+{action: "help"}
+
+// Get comprehensive examples for task tool
+{action: "example"}
+```
+
+### For AI Agents (68% Token Reduction)
+
+**Tool Selection & Workflows:**
+- 📖 **[Tool Selection](docs/TOOL_SELECTION.md)** - Decision tree, when to use each tool (236 lines, ~12k tokens)
+- 🔄 **[Workflows](docs/WORKFLOWS.md)** - Multi-step examples, multi-agent coordination (602 lines, ~30k tokens)
+- 📚 **[Tool Reference](docs/TOOL_REFERENCE.md)** - Parameters, batch operations, templates (471 lines, ~24k tokens)
+- ✅ **[Best Practices](docs/BEST_PRACTICES.md)** - Common errors, troubleshooting (345 lines, ~17k tokens)
+
+**Task System:**
+- 📋 **[Task Overview](docs/TASK_OVERVIEW.md)** - Lifecycle, status transitions, auto-stale (363 lines, ~10k tokens)
+- ⚙️ **[Task Actions](docs/TASK_ACTIONS.md)** - All actions with examples (854 lines, ~21k tokens)
+- 🔗 **[Task Linking](docs/TASK_LINKING.md)** - Link tasks to decisions/constraints/files (729 lines, ~18k tokens)
+- 🔄 **[Task Migration](docs/TASK_MIGRATION.md)** - Migrate from decision-based tracking (701 lines, ~18k tokens)
+
+**Shared References:**
+- 📘 **[Shared Concepts](docs/SHARED_CONCEPTS.md)** - Layer definitions, enum values, atomic mode (339 lines, ~17k tokens)
+- 🏗️ **[Architecture](docs/ARCHITECTURE.md)** - Technical architecture and database schema
+
+### For Developers
+
+- **[Building from Source](docs/ARCHITECTURE.md#development)** - Setup and build instructions
+- **[Migration Guides](docs/MIGRATION_v2.md)** - Version upgrade guides
+- **[CLI Tool](docs/AI_AGENT_GUIDE.md#cli-usage)** - Query database from terminal
+
+## Examples
+
+### Multi-Agent Coordination
+
+```javascript
+// Orchestrator creates tasks
+{
+  action: "batch_create",
+  tasks: [
+    {title: "Setup database", assigned_agent: "db-agent"},
+    {title: "Create API", assigned_agent: "api-agent"}
+  ]
+}
+
+// Agents send status updates
+{
+  action: "send",
+  from_agent: "db-agent",
+  to_agent: "orchestrator",
+  message: "Database ready",
+  priority: "high"
+}
+```
+
+### Breaking Change Management
+
+```javascript
+// Record deprecation
+{
+  action: "set",
+  key: "api_v1_deprecated",
+  value: "/v1 endpoints deprecated, use /v2",
+  tags: ["breaking-change", "api"]
+}
+
+// Add constraint
+{
+  action: "add",
+  category: "architecture",
+  constraint_text: "All endpoints must use /v2 prefix",
+  priority: "high"
+}
+```
+
+### Session Continuity
+
+```javascript
+// Session 1: Save state
+{
+  action: "set",
+  key: "refactor_progress",
+  value: "Completed 3/5 modules",
+  tags: ["session-state"]
+}
+
+// Session 2: Resume work
+{
+  action: "get",
+  key: "refactor_progress"
+}
+// → Returns: "Completed 3/5 modules"
+```
+
+## Performance
+
+- **Query speed**: 2-50ms
+- **Concurrent agents**: 5+ simultaneous
+- **Storage efficiency**: ~140 bytes per decision
+- **Token savings**: 60-75% in typical projects
+
+## Version
+
+Current version: **3.0.0**
+See [CHANGELOG.md](CHANGELOG.md) for release history.
+
+## License
+
+MIT - see [LICENSE](LICENSE) file
+
+## Links
+
+- [npm package](https://www.npmjs.com/package/sqlew)
+- [GitHub repository](https://github.com/sin5ddd/mcp-sqlew)
+- [Model Context Protocol](https://modelcontextprotocol.io/)
+
+## Support
+
+- **Issues**: [GitHub Issues](https://github.com/sin5ddd/mcp-sqlew/issues)
+- **Documentation**: [docs/](docs/) directory
+
+## Acknowledgments
+
+Built with [Model Context Protocol SDK](https://github.com/modelcontextprotocol/sdk), [better-sqlite3](https://github.com/WiseLibs/better-sqlite3), and TypeScript.
+
+**Author**: sin5ddd