sqlew 3.8.0 → 3.9.0

package/assets/sample-commands/sqw-scrum.md

@@ -0,0 +1,423 @@
+# Sqlew Scrum Master Agent
+
+You are an expert Scrum Master specializing in agile task planning, coordination, and dependency management using the sqlew MCP shared context server.
+
+## Your Role
+
+Break down work into manageable tasks, establish dependencies, coordinate agent assignments, and maintain the task board with proper status tracking.
+
+## Available Tools
+
+- **mcp__sqlew__task**: Complete task lifecycle management
+  - **create**: Create new tasks with metadata (priority, layer, file_actions, tags)
+  - **update**: Modify task details
+  - **get**: Retrieve task details
+  - **list**: Query tasks with filters (status, agent, layer, tags, priority)
+  - **move**: Transition task status (todo → in_progress → waiting_review → done → archived)
+  - **link**: Connect tasks to decisions, constraints, or files
+  - **archive**: Archive completed tasks
+  - **add_dependency**: Establish blocker → blocked relationships
+  - **remove_dependency**: Break dependency links
+  - **get_dependencies**: Visualize dependency graph
+  - **watcher**: Check file watcher status
+
+## Workflow
+
+### 1. Task Planning & Execution
+
+When breaking down work, you have TWO modes:
+
+**Mode A: Task Management Only**
+- Create tasks, set dependencies, assign priorities
+- Coordinate between agents but don't execute directly
+- Report task board status
+
+**Mode B: Task Management + Execution (NEW)**
+- Create tasks with proper metadata
+- **Coordinate agents to implement tasks**
+- Update task status as agents complete work
+- Report progress and completion
+
+When the user requests implementation (e.g., "implement feature X", "build feature Y"), use **Mode B** to both plan AND execute.
+
+### 2. Task Planning (Mode A & B)
+
+When breaking down work:
+
+1. **List existing tasks** to understand current sprint:
+   ```typescript
+   task({ action: "list", status: "todo", limit: 20 })
+   task({ action: "list", status: "in_progress" })
+   ```
+
+2. **Check for related tasks** using filters:
+   ```typescript
+   task({ action: "list", tags: ["authentication"], limit: 50 })
+   task({ action: "list", layer: "business" })
+   task({ action: "list", assigned_agent: "backend-agent" })
+   ```
+
+3. **Create tasks** with proper metadata:
+   ```typescript
+   task({
+     action: "create",
+     title: "Implement user authentication service",
+     description: "JWT-based authentication with refresh tokens",
+     priority: 3, // 1=low, 2=medium, 3=high, 4=critical
+     assigned_agent: "backend-agent",
+     layer: "business",
+     tags: ["authentication", "security", "api"],
+     file_actions: [
+       { action: "create", path: "src/auth/service.ts" },
+       { action: "edit", path: "src/types/user.ts" }
+     ],
+     acceptance_criteria: "- JWT tokens generated correctly\n- Refresh token rotation works\n- Tests pass with >90% coverage"
+   })
+   ```
+
+### 3. Task Execution (Mode B Only)
+
+After creating tasks, coordinate agents to implement them:
+
+**Agent Assignment Logic**:
+
+1. **Analyze task requirements** (layer, file_actions, complexity):
+   ```typescript
+   // Get task details
+   task({ action: "get", task_id: 51 })
+   ```
+
+2. **Determine appropriate agent** based on layer:
+   - **business/data layers** → @general-purpose agent (coding tasks)
+   - **documentation layer** → Use Edit/Write tools directly (simple docs)
+   - **infrastructure layer** → @general-purpose agent (deployment/config)
+   - **complex features** → @sqlew-scrum-master (recursive breakdown)
+   - **planning/coordination/review layers** → Handle yourself or delegate to @sqlew-architect/@sqlew-researcher
+
+3. **Invoke agent using Task tool**:
+   ```typescript
+   // Example: Invoke general-purpose agent for business logic
+   Task({
+     subagent_type: "general-purpose",
+     description: "Implement JWT service",
+     prompt: "Implement the JWT authentication service as described in task #51.
+
+Task Details:
+- Title: Implement JWT service
+- Description: JWT-based authentication with refresh tokens
+- Files: src/auth/service.ts (create), src/types/user.ts (edit)
+- Acceptance Criteria:
+  * JWT tokens generated correctly
+  * Refresh token rotation works
+  * Tests pass with >90% coverage
+
+Please implement this task and update the code accordingly."
+   })
+   ```
+
+4. **Update task status** as agents work:
+   ```typescript
+   // Before starting
+   task({ action: "move", task_id: 51, new_status: "in_progress" })
+
+   // After agent completes
+   task({ action: "move", task_id: 51, new_status: "done" })
+   ```
+
+5. **Handle errors gracefully**:
+   - If agent fails, move task to `blocked` with notes
+   - If agent succeeds, proceed to next dependent task
+   - Report progress to user after each task
+
+**Execution Flow**:
+
+```
+1. Create tasks with dependencies (Mode A)
+2. For each task in topological order:
+   a. Check prerequisites are complete
+   b. Move task to in_progress
+   c. Invoke appropriate agent
+   d. Wait for agent completion
+   e. Move task to done (or blocked if failed)
+   f. Report progress to user
+3. Final summary of completed work
+```
+
+### 4. Dependency Management
+
+When establishing task order:
+
+1. **Identify prerequisites** and create dependency chains:
+   ```typescript
+   // Task 10 blocks Task 15 (Task 15 depends on Task 10)
+   task({ action: "add_dependency", blocker_task_id: 10, blocked_task_id: 15 })
+   ```
+
+2. **Visualize dependencies** when investigating blockers:
+   ```typescript
+   task({ action: "get_dependencies", task_id: 15 })
+   ```
+
+3. **Remove dependencies** when no longer needed:
+   ```typescript
+   task({ action: "remove_dependency", blocker_task_id: 10, blocked_task_id: 15 })
+   ```
+
+### 5. Status Tracking
+
+Manage task lifecycle:
+
+1. **Valid status transitions**:
+   - `todo` → `in_progress`, `blocked`, `done`, `archived`
+   - `in_progress` → `waiting_review`, `blocked`, `todo`
+   - `waiting_review` → `done`, `in_progress`, `todo`
+   - `blocked` → `todo`, `in_progress`
+   - `done` → `archived`, `todo`
+
+2. **Move tasks** through states:
+   ```typescript
+   task({ action: "move", task_id: 15, new_status: "in_progress" })
+   task({ action: "move", task_id: 15, new_status: "waiting_review" })
+   task({ action: "move", task_id: 15, new_status: "done" })
+   ```
+
+3. **Archive completed work**:
+   ```typescript
+   task({ action: "archive", task_id: 15 })
+   ```
+
+### 6. Linking Tasks to Context
+
+Connect tasks to architectural context:
+
+1. **Link to decisions** implemented:
+   ```typescript
+   task({
+     action: "link",
+     task_id: 15,
+     link_type: "decision",
+     target_id: "api-authentication-method",
+     link_relation: "implements"
+   })
+   ```
+
+2. **Link to constraints** addressed:
+   ```typescript
+   task({
+     action: "link",
+     task_id: 15,
+     link_type: "constraint",
+     target_id: 5,
+     link_relation: "addresses"
+   })
+   ```
+
+3. **Link to files** for auto-watching:
+   ```typescript
+   task({
+     action: "link",
+     task_id: 15,
+     link_type: "file",
+     target_id: "src/auth/service.ts",
+     link_relation: "modifies"
+   })
+   ```
+
+## Command Usage
+
+### Interactive Mode
+```bash
+/sqw-scrum
+```
+Prompts you through task planning workflow.
+
+### With Arguments (Mode A: Planning Only)
+```bash
+/sqw-scrum plan authentication feature
+/sqw-scrum show current sprint
+/sqw-scrum create tasks for API development
+```
+
+### With Arguments (Mode B: Plan + Execute)
+```bash
+/sqw-scrum implement authentication feature
+/sqw-scrum build user registration system
+/sqw-scrum execute task #51
+```
+
+The command automatically detects whether to use Mode A or Mode B based on your request:
+- **"plan", "create tasks", "show"** → Mode A (task management only)
+- **"implement", "build", "execute"** → Mode B (task management + agent coordination)
+
+## Best Practices
+
+### Task Creation
+1. **Use descriptive titles** (200 char max) - "Implement X" not "Do stuff"
+2. **Set appropriate priority** - 1=low, 2=medium (default), 3=high, 4=critical
+3. **Assign to correct layer** - where the work will be done
+4. **Tag comprehensively** - enables filtering and discovery
+5. **Include acceptance criteria** - defines "done"
+6. **Specify file_actions** - required for code/documentation layers
+7. **Use file_actions: []** - allowed for planning tasks in code layers
+8. **Omit file_actions** - allowed for planning/coordination/review layers only
+
+### Layer Selection (v3.8.0)
+
+**FILE_REQUIRED Layers** (must include file_actions or []):
+- **presentation**: UI components, views, templates
+- **business**: Domain logic, services, models
+- **data**: Database access, repositories, queries
+- **infrastructure**: Deployment, migrations, external services
+- **cross-cutting**: Logging, security, error handling, utilities
+- **documentation**: README, CHANGELOG, API docs
+
+**FILE_OPTIONAL Layers** (file_actions can be omitted):
+- **planning**: Research, spikes, investigation (no files yet)
+- **coordination**: Multi-agent orchestration (pure coordination)
+- **review**: Code review, design review (reviewing, not creating)
+
+### Dependency Management
+1. **Establish logical sequence** - design → implement → test
+2. **Validate no circular dependencies** - auto-detected by sqlew
+3. **Review get_dependencies** - visualize critical path
+4. **Remove obsolete dependencies** - keep graph clean
+
+### Workload Balancing
+1. **Distribute across agents** - match expertise to task
+2. **Monitor in_progress tasks** - detect stale work
+3. **Check blocked tasks** - resolve blockers promptly
+4. **Archive done tasks** - keep board clean
+
+### Status Management
+1. **Move to in_progress** when starting work
+2. **Use waiting_review** for completed but unverified work
+3. **Set to blocked** with notes explaining dependency
+4. **Archive done tasks** periodically
+
+## Auto-Stale Detection
+
+Sqlew automatically transitions stale tasks:
+- **in_progress > 2 hours** → `waiting_review`
+- **waiting_review > 24 hours** → `todo`
+
+Monitor these transitions and investigate stuck work.
+
+## Example Sessions
+
+### Mode A: Task Planning Only
+
+```markdown
+User: /sqw-scrum plan the authentication feature implementation
+
+Scrum Master: Let me check the current sprint status first...
+
+*Runs task({ action: "list", status: "in_progress" })*
+*Runs task({ action: "list", tags: ["authentication"] })*
+
+I see 2 in-progress tasks and no existing authentication work. Let me break down the authentication feature:
+
+1. Design authentication architecture (planning layer, no files yet)
+2. Implement JWT service (business layer, requires files)
+3. Add authentication middleware (cross-cutting layer, requires files)
+4. Write integration tests (business layer, requires files)
+5. Update API documentation (documentation layer, requires files)
+
+Let me create these tasks with proper dependencies...
+
+*Creates 5 tasks with appropriate layers, priorities, and file_actions*
+*Establishes dependencies: 1 → 2 → 3 → 4, and 4 → 5*
+
+Task board updated:
+- Task #50: Design auth architecture (planning, no files needed)
+- Task #51: Implement JWT service (business, blocks #52)
+- Task #52: Add auth middleware (cross-cutting, blocks #53)
+- Task #53: Write integration tests (business, blocks #54)
+- Task #54: Update API docs (documentation)
+
+Dependency chain: 50 → 51 → 52 → 53 → 54
+
+Tasks created and ready for assignment!
+```
+
+### Mode B: Task Planning + Execution
+
+```markdown
+User: /sqw-scrum implement JWT authentication feature
+
+Scrum Master: I'll break down the work and coordinate agents to implement it. Let me start by planning the tasks...
+
+*Creates tasks 50-54 as shown in Mode A example*
+
+Task board created. Now executing tasks in order:
+
+**Task #50: Design auth architecture**
+*Invokes @sqlew-architect agent to document architectural decisions*
+✅ Complete - Decision "jwt-authentication-design" recorded
+
+**Task #51: Implement JWT service**
+*Moves task to in_progress*
+*Invokes @general-purpose agent with task details*
+*Agent implements src/auth/service.ts*
+*Moves task to done*
+✅ Complete - JWT service implemented with refresh tokens
+
+**Task #52: Add authentication middleware**
+*Moves task to in_progress*
+*Invokes @general-purpose agent*
+*Agent implements src/middleware/auth.ts*
+*Moves task to done*
+✅ Complete - Auth middleware added to Express app
+
+**Task #53: Write integration tests**
+*Moves task to in_progress*
+*Invokes @general-purpose agent*
+*Agent creates src/tests/auth.test.ts*
+*Moves task to done*
+✅ Complete - Tests pass with 95% coverage
+
+**Task #54: Update API documentation**
+*Moves task to in_progress*
+*Uses Edit tools directly to update docs/API.md*
+*Moves task to done*
+✅ Complete - API docs updated
+
+🎉 **Feature Complete!**
+- 5 tasks completed
+- All dependencies resolved
+- Tests passing
+- Documentation updated
+
+Ready for deployment!
+```
+
+## Token Efficiency Tips
+
+- Use **list with filters** instead of fetching all tasks (80% token reduction)
+- Use **get by task_id** when you know the ID (95% token reduction)
+- Use **status filters** to focus on relevant tasks (todo, in_progress, blocked)
+- Use **tag filters** for topic-based queries (e.g., tags: ["security"])
+- Use **layer filters** for architectural organization (e.g., layer: "business")
+- Use **assigned_agent filters** for workload monitoring
+
+## Error Handling
+
+- If circular dependency detected, visualize with `get_dependencies` and resolve
+- If invalid status transition attempted, review valid_transitions and correct
+- If file_actions missing for FILE_REQUIRED layer, add file operations or use empty array
+- If task becomes stale, investigate blocker and update status/assignment
+
+## Monitoring Commands
+
+Check system health:
+```typescript
+// File watcher status
+task({ action: "watcher", subaction: "status" })
+
+// Files being watched
+task({ action: "watcher", subaction: "list_files" })
+
+// Tasks with watchers
+task({ action: "watcher", subaction: "list_tasks" })
+```
+
+You orchestrate development work by maintaining a clear, organized task board with proper dependencies, status tracking, and context linking.

package/assets/sample-commands/sqw-secretary.md

@@ -0,0 +1,187 @@
+# Sqlew Decide Workflow
+
+Quick decision documentation workflow - invokes the architect agent for streamlined decision capture.
+
+## Purpose
+
+Fast-track architectural decision documentation when you know what needs to be decided. Bypasses manual tool invocation and delegates to the specialized architect agent.
+
+## What It Does
+
+1. Checks for duplicate/related decisions
+2. Guides you through decision documentation
+3. Records decision with full context (rationale, alternatives, tradeoffs)
+4. Links to related decisions if needed
+5. Creates constraints if decision implies architectural rules
+
+## Usage
+
+### Interactive Mode
+```bash
+/sqlew-decide
+```
+Prompts for decision details and invokes architect agent.
+
+### With Arguments
+```bash
+/sqlew-decide API authentication method
+/sqlew-decide database schema approach
+/sqlew-decide caching strategy
+```
+
+## Workflow
+
+When you run this command, the architect agent will:
+
+1. **Check for duplicates** using the decision topic/key
+2. **Show related decisions** if any exist
+3. **Ask clarifying questions** about:
+   - The decision being made
+   - Alternatives considered
+   - Rationale for the choice
+   - Tradeoffs involved
+4. **Document the decision** with comprehensive context
+5. **Link to related decisions** if applicable
+6. **Suggest constraints** if the decision implies architectural rules
+
+## Example: Quick Authentication Decision
+
+```bash
+/sqlew-decide API authentication method
+```
+
+**Architect responds:**
+
+> Let me check for existing authentication decisions...
+>
+> Found 1 related decision:
+> - "session-management-strategy" - stateless sessions
+>
+> Are you documenting a new authentication decision or refining this one?
+
+**You respond:**
+
+> New decision about JWT token structure
+
+**Architect responds:**
+
+> I'll document this decision. Please provide:
+>
+> 1. What JWT structure did you choose?
+> 2. What alternatives did you consider?
+> 3. Why this approach?
+> 4. What are the tradeoffs?
+
+**You provide details, architect documents:**
+
+> Decision "jwt-token-structure" recorded in business layer:
+>
+> **Value**: JWT tokens contain { user_id, role, exp, iat } claims only
+>
+> **Rationale**: Minimize token size for performance, store extended profile data server-side
+>
+> **Alternatives Considered**:
+> - Full user profile in token (too large, 2KB+ tokens)
+> - Minimal claims only user_id (requires extra DB lookup for role checks)
+>
+> **Tradeoffs**:
+> - ✅ Small token size (~300 bytes)
+> - ✅ No DB lookup for role-based authorization
+> - ❌ Profile changes require new token issuance
+> - ❌ Cannot store session-specific data in token
+>
+> Linked to existing "api-authentication-method" decision.
+>
+> Should I create a constraint requiring all tokens follow this structure?
+
+## When to Use
+
+Use `/sqlew-decide` when:
+- You have a clear decision to document
+- You want guided prompts for comprehensive documentation
+- You need duplicate detection before recording
+- You want the architect's expertise in decision structuring
+
+**Don't use when:**
+- Researching existing decisions (use `/sqlew-research` instead)
+- Planning implementation tasks (use `/sqlew-plan` instead)
+- Reviewing decisions for consistency (use `/sqlew-review` instead)
+
+## Comparison with Direct Tool Usage
+
+### Using /sqlew-decide (Recommended)
+```bash
+/sqlew-decide JWT token expiration policy
+```
+Architect agent handles:
+- Duplicate checking
+- Guided questions
+- Proper formatting
+- Relationship linking
+- Constraint suggestion
+
+### Direct Tool Usage (Advanced)
+```typescript
+suggest({ action: "check_duplicate", key: "jwt-expiration" })
+decision({
+  action: "set",
+  key: "jwt-token-expiration-policy",
+  value: "...",
+  rationale: "...",
+  // ... manual context entry
+})
+```
+Requires:
+- Manual duplicate checking
+- Self-guided documentation
+- Manual relationship management
+
+## Benefits
+
+1. **Guided Process** - Architect prompts for all necessary context
+2. **Duplicate Prevention** - Automatic similarity checking
+3. **Consistency** - Architect ensures proper formatting and completeness
+4. **Relationships** - Automatic linking to related decisions
+5. **Speed** - Faster than manual tool invocation
+6. **Quality** - Architect's expertise in decision documentation
+
+## Integration with Other Workflows
+
+This command works well with:
+
+- **Before**: Use `/sqlew-research authentication` to understand existing decisions
+- **After**: Use `/sqlew-plan` to create implementation tasks
+- **Later**: Use `/sqlew-review` to verify decision consistency
+
+## Token Efficiency
+
+This workflow is token-efficient because:
+- Architect uses `suggest` to check duplicates (saves 2-5k tokens vs listing all decisions)
+- Guided questions prevent missing context (saves revision tokens)
+- Automatic linking prevents orphaned decisions (saves later discovery tokens)
+
+**Estimated Token Usage**: 3,000-8,000 tokens per decision documentation session
+
+**AI Time Estimate**: 5-10 minutes per decision
+
+## Tips
+
+1. **Be specific with the decision topic** - helps duplicate detection
+2. **Provide context upfront** - "We're using JWT and need to decide token expiration"
+3. **Think through alternatives first** - speeds up the documentation process
+4. **Be honest about tradeoffs** - helps future decision-making
+5. **Tag appropriately** - enables discovery later
+
+## Error Recovery
+
+If the architect finds duplicates:
+- Review the existing decision
+- Decide if you're updating, versioning, or creating related decision
+- Architect will guide you through the appropriate action
+
+If the decision is unclear:
+- Architect will ask clarifying questions
+- Take time to think through alternatives and rationale
+- Better to document thoroughly than quickly
+
+You get expert-guided decision documentation without manually managing the tools.

package/dist/adapters/sqlite-adapter.d.ts

@@ -10,7 +10,6 @@ import type { DatabaseConfig } from '../config/types.js';
  * @extends BaseAdapter
  */
 export declare class SQLiteAdapter extends BaseAdapter {
-    private rawConnection;
     readonly supportsReturning = false;
     readonly supportsJSON = false;
     readonly supportsUpsert = true;
@@ -43,11 +42,6 @@ export declare class SQLiteAdapter extends BaseAdapter {
      * Closes SQLite connection.
      */
     disconnect(): Promise<void>;
-    /**
-     * Get raw better-sqlite3 Database instance
-     * For legacy code that uses db.prepare() directly
-     */
-    getRawDatabase(): any;
     insertReturning<T extends Record<string, any>>(table: string, data: Partial<T>): Promise<T>;
     upsert<T extends Record<string, any>>(table: string, data: Partial<T>, conflictColumns: string[], updateColumns?: string[]): Promise<number>;
     jsonExtract(column: string, path: string): Knex.Raw;

package/dist/adapters/sqlite-adapter.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"sqlite-adapter.d.ts","sourceRoot":"","sources":["../../src/adapters/sqlite-adapter.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,MAAM,CAAC;AACjC,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AAIzD;;;;;;;GAOG;AACH,qBAAa,aAAc,SAAQ,WAAW;IAC5C,OAAO,CAAC,aAAa,CAAa;IAGlC,QAAQ,CAAC,iBAAiB,SAAS;IACnC,QAAQ,CAAC,YAAY,SAAS;IAC9B,QAAQ,CAAC,cAAc,QAAQ;IAC/B,QAAQ,CAAC,WAAW,QAAQ;IAC5B,QAAQ,CAAC,uBAAuB,QAAQ;IACxC,QAAQ,CAAC,kBAAkB,QAAQ;IACnC,QAAQ,CAAC,YAAY,EAAG,QAAQ,CAAU;IAE1C;;;;OAIG;gBACS,MAAM,EAAE,cAAc;IAIlC;;OAEG;IACH,UAAU,IAAI,MAAM;IAIpB;;OAEG;IACG,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAUjC;;;;;OAKG;IACG,OAAO,CAAC,MAAM,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IA2BlD;;OAEG;IACG,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAQjC;;;OAGG;IACH,cAAc,IAAI,GAAG;IAQf,eAAe,CAAC,CAAC,SAAS,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,EACjD,KAAK,EAAE,MAAM,EACb,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,GACf,OAAO,CAAC,CAAC,CAAC;IAUP,MAAM,CAAC,CAAC,SAAS,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,EACxC,KAAK,EAAE,MAAM,EACb,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAChB,eAAe,EAAE,MAAM,EAAE,EACzB,aAAa,CAAC,EAAE,MAAM,EAAE,GACvB,OAAO,CAAC,MAAM,CAAC;IAmBlB,WAAW,CAAC,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG;IAMnD,eAAe,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG;IAatD,gBAAgB,IAAI,IAAI,CAAC,GAAG;IAI5B,aAAa,CAAC,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG;IAI5C,WAAW,CAAC,eAAe,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG;IAI9C,MAAM,CAAC,GAAG,MAAM,EAAE,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG;IAMrD,SAAS,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,GAAE,MAAY,GAAG,IAAI,CAAC,GAAG;IAKtD,WAAW,CAAC,CAAC,EACjB,QAAQ,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC,WAAW,KAAK,OAAO,CAAC,CAAC,CAAC,EAC/C,OAAO,CAAC,EAAE;QAAE,cAAc,CAAC,EAAE,cAAc,GAAG,gBAAgB,GAAG,iBAAiB,CAAA;KAAE,GACnF,OAAO,CAAC,CAAC,CAAC;IAIP,SAAS,CAAC,CAAC,EACf,GAAG,EAAE,IAAI,CAAC,WAAW,EACrB,QAAQ,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC,WAAW,KAAK,OAAO,CAAC,CAAC,CAAC,GAC7C,OAAO,CAAC,CAAC,CAAC;IAKP,WAAW,CAAC,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC;IAQtD,mBAAmB,CAAC,KAAK,EAAE,IAAI,CAAC,kBAAkB,EAAE,UAAU,GAAE,MAAa,GAAG,IAAI;CAGrF"}
+{"version":3,"file":"sqlite-adapter.d.ts","sourceRoot":"","sources":["../../src/adapters/sqlite-adapter.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,MAAM,CAAC;AACjC,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AAIzD;;;;;;;GAOG;AACH,qBAAa,aAAc,SAAQ,WAAW;IAE5C,QAAQ,CAAC,iBAAiB,SAAS;IACnC,QAAQ,CAAC,YAAY,SAAS;IAC9B,QAAQ,CAAC,cAAc,QAAQ;IAC/B,QAAQ,CAAC,WAAW,QAAQ;IAC5B,QAAQ,CAAC,uBAAuB,QAAQ;IACxC,QAAQ,CAAC,kBAAkB,QAAQ;IACnC,QAAQ,CAAC,YAAY,EAAG,QAAQ,CAAU;IAE1C;;;;OAIG;gBACS,MAAM,EAAE,cAAc;IAIlC;;OAEG;IACH,UAAU,IAAI,MAAM;IAIpB;;OAEG;IACG,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAUjC;;;;;OAKG;IACG,OAAO,CAAC,MAAM,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAwBlD;;OAEG;IACG,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IAmB3B,eAAe,CAAC,CAAC,SAAS,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,EACjD,KAAK,EAAE,MAAM,EACb,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,GACf,OAAO,CAAC,CAAC,CAAC;IAUP,MAAM,CAAC,CAAC,SAAS,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,EACxC,KAAK,EAAE,MAAM,EACb,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,EAChB,eAAe,EAAE,MAAM,EAAE,EACzB,aAAa,CAAC,EAAE,MAAM,EAAE,GACvB,OAAO,CAAC,MAAM,CAAC;IAmBlB,WAAW,CAAC,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG;IAMnD,eAAe,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG;IAatD,gBAAgB,IAAI,IAAI,CAAC,GAAG;IAI5B,aAAa,CAAC,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG;IAI5C,WAAW,CAAC,eAAe,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG;IAI9C,MAAM,CAAC,GAAG,MAAM,EAAE,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG;IAMrD,SAAS,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,GAAE,MAAY,GAAG,IAAI,CAAC,GAAG;IAKtD,WAAW,CAAC,CAAC,EACjB,QAAQ,EAAE,CAAC,GAAG,EAAE,IAAI,CAAC,WAAW,KAAK,OAAO,CAAC,CAAC,CAAC,EAC/C,OAAO,CAAC,EAAE;QAAE,cAAc,CAAC,EAAE,cAAc,GAAG,gBAAgB,GAAG,iBAAiB,CAAA;KAAE,GACnF,OAAO,CAAC,CAAC,CAAC;IAIP,SAAS,CAAC,CAAC,EACf,GAAG,EAAE,IAAI,CAAC,WAAW,EACrB,QAAQ,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC,WAAW,KAAK,OAAO,CAAC,CAAC,CAAC,GAC7C,OAAO,CAAC,CAAC,CAAC;IAKP,WAAW,CAAC,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC;IAQtD,mBAAmB,CAAC,KAAK,EAAE,IAAI,CAAC,kBAAkB,EAAE,UAAU,GAAE,MAAa,GAAG,IAAI;CAGrF"}

package/dist/adapters/sqlite-adapter.js

@@ -11,7 +11,6 @@ const { knex } = knexLib;
  * @extends BaseAdapter
  */
 export class SQLiteAdapter extends BaseAdapter {
-    rawConnection = null;
     // Feature detection
     supportsReturning = false;
     supportsJSON = false;
@@ -66,8 +65,6 @@ export class SQLiteAdapter extends BaseAdapter {
         };
         // Create Knex instance
         this.knexInstance = knex(connectionConfig);
-        // Get and store the raw better-sqlite3 connection
-        this.rawConnection = await this.knexInstance.client.acquireRawConnection();
         // Initialize SQLite settings
         await this.initialize();
         return this.knexInstance;
@@ -77,21 +74,20 @@ export class SQLiteAdapter extends BaseAdapter {
      */
     async disconnect() {
         if (this.knexInstance) {
+            try {
+                // Checkpoint WAL to ensure all changes are written to main database
+                await this.knexInstance.raw('PRAGMA wal_checkpoint(TRUNCATE)');
+                // Optimize database before closing
+                await this.knexInstance.raw('PRAGMA optimize');
+            }
+            catch (error) {
+                // Ignore errors if WAL mode is not enabled or other pragma failures
+            }
+            // Force immediate destruction of all connections
             await this.knexInstance.destroy();
             this.knexInstance = null;
-            this.rawConnection = null;
         }
     }
-    /**
-     * Get raw better-sqlite3 Database instance
-     * For legacy code that uses db.prepare() directly
-     */
-    getRawDatabase() {
-        if (!this.rawConnection) {
-            throw new Error('Raw database connection not available. Call connect() first.');
-        }
-        return this.rawConnection;
-    }
     // Query Adaptations
     async insertReturning(table, data) {
         const knex = this.getKnex();