Loading ToolBench
Visual indicator and management UI for MCP Memory Bank custom modes. A VSCode extension that provides status bar indicators, side panel management, and file watchers for tracking development modes (VAN, PLAN, CREATIVE, IMPLEMENT, SECURITY, QA, REFLECT) and tasks.
This is a VS Code extension, not an MCP server. The source code shows VS Code commands and file I/O operations, not MCP tool definitions with proper schemas and descriptions. The package.json declares 5 'commands' in VS Code's contributes.commands array, which is a VS Code UI pattern, NOT MCP tool registration. No MCP protocol implementation is visible, no tool schemas, no JSON-RPC handlers, no transport layer. The tool names are prefixed with 'mcp-memory-bank.' but this is a VS Code command namespace, not MCP tool naming. No input parameter schemas are visible anywhere in the source. Descriptions exist in the package.json 'title' fields but are extremely brief (10-20 chars) and lack parameter documentation. Critical issue: this appears to be a VS Code sidebar extension for managing a local Markdown-based task system, not an MCP server that can be called by agents or integrated into MCP clients.
Initialize Memory Bank directory structure with default tasks.md and activeContext.md files if they don't exist
Open task documentation directory and README.md file for the current active task
Switch the current MCP Memory Bank mode via quick pick selection from available modes (VAN, PLAN, CREATIVE, IMPLEMENT, SECURITY, QA, REFLECT)
Update the active context (placeholder for Phase 2 implementation)
Open and display the tasks.md file containing active tasks and their metadata
Not an MCP server, this is a VS Code extension. No MCP protocol implementation detected. The 'tools' are VS Code commands, not MCP-registered tools with schemas and proper tool definitions.
No input parameter schemas visible. Tools accept arguments (e.g., switchMode takes 'workspaceRoot' and 'currentMode') but no JSON Schema is defined, no type constraints, no parameter descriptions.
Descriptions in package.json are extremely brief (10-20 chars). 'Switch the current MCP Memory Bank mode via quick pick selection...' is understandable but lacks parameter detail. Tool descriptions should be 50 - 200 chars, with a WHAT/WHEN/HOW breakdown for LLM selection.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 21 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 8 | - | v1 |
Naming uses dot notation (mcp-memory-bank.switchMode) typical of VS Code command IDs, not verb_noun MCP convention. MCP tools should use names like 'switch_memory_bank_mode', 'initialize_memory_bank', 'get_tasks', etc.
Error handling returns vscode.window messages (UI dialogs) rather than structured error objects. MCP tools must return structured errors with error codes, messages, and recovery hints for LLM consumption, not UI popup strings.
No output schema documentation. Each command returns Promise<void> or opens a VS Code window, with no structured result data for downstream tool chaining or agent reasoning.