An AI-powered task-management system you can drop into Cursor, Lovable, Windsurf, Roo, and others.
JavaScript aicursorcursor-aicursorailovablelovable-devroocodetask-managertaskstasks-listwindsurfwindsurf-ai
feat: change default MCP tool tier from 'all' to 'core' (#1490)
Ralph Khreishbf4515b6de28
diff --git a/.taskmaster/CLAUDE.md b/.taskmaster/CLAUDE.mdindex 4f030f9c..a625746e 100644--- a/.taskmaster/CLAUDE.md+++ b/.taskmaster/CLAUDE.md@@ -91,6 +91,7 @@ Task Master provides an MCP server that Claude Code can connect to. Configure in"command": "npx","args": ["-y", "task-master-ai"],"env": {+ "TASK_MASTER_TOOLS": "core","ANTHROPIC_API_KEY": "your_key_here","PERPLEXITY_API_KEY": "your_key_here","OPENAI_API_KEY": "OPENAI_API_KEY_HERE",@@ -106,6 +107,24 @@ Task Master provides an MCP server that Claude Code can connect to. Configure in}```+### MCP Tool Tiers++Task Master uses tiered tool loading to optimize context window usage:++| Tier | Tools | Use Case |+|------|-------|----------|+| `core` | 7 | Minimal daily workflow tools (default) |+| `standard` | 14 | Common task management |+| `all` | 44+ | Full suite with research, autopilot, dependencies |++**Core tools (7):** `get_tasks`, `next_task`, `get_task`, `set_task_status`, `update_subtask`, `parse_prd`, `expand_task`++**Standard adds (7 more):** `initialize_project`, `analyze_project_complexity`, `expand_all`, `add_subtask`, `remove_task`, `add_task`, `complexity_report`++**All tier adds:** Dependency management, tag management, research, autopilot TDD workflow, scope up/down, models, rules++**To upgrade tiers:** Change `TASK_MASTER_TOOLS` in `.mcp.json` from `"core"` to `"standard"` or `"all"` and restart the MCP connection.+### Essential MCP Tools```javascriptdiff --git a/apps/docs/capabilities/mcp.mdx b/apps/docs/capabilities/mcp.mdxindex 1c210863..d748f36d 100644--- a/apps/docs/capabilities/mcp.mdx+++ b/apps/docs/capabilities/mcp.mdx@@ -19,27 +19,7 @@ To optimize LLM context usage, you can control which Task Master MCP tools are l### Configuration Modes-#### All Tools (Default)-Loads all 36 available tools. Use when you need full Task Master functionality.--```json-{- "mcpServers": {- "task-master-ai": {- "command": "npx",- "args": ["-y", "task-master-ai"],- "env": {- "TASK_MASTER_TOOLS": "all",- "ANTHROPIC_API_KEY": "your_key_here"- }- }- }-}-```--If `TASK_MASTER_TOOLS` is not set, all tools are loaded by default.--#### Core Tools (Lean Mode)+#### Core Tools (Default, Lean Mode)Loads only 7 essential tools for daily development. Ideal for minimal context usage.**Core tools included:**@@ -66,7 +46,7 @@ Loads only 7 essential tools for daily development. Ideal for minimal context us}```-You can also use `"lean"` as an alias for `"core"`.+You can also use `"lean"` as an alias for `"core"`. If `TASK_MASTER_TOOLS` is not set, core tools are loaded by default.#### Standard ToolsLoads 15 commonly used tools. Balances functionality with context efficiency.@@ -114,9 +94,27 @@ Specify exactly which tools to load using a comma-separated list. Tool names are}```+#### All Tools+Loads all 36 available tools. Use when you need full Task Master functionality.++```json+{+ "mcpServers": {+ "task-master-ai": {+ "command": "npx",+ "args": ["-y", "task-master-ai"],+ "env": {+ "TASK_MASTER_TOOLS": "all",+ "ANTHROPIC_API_KEY": "your_key_here"+ }+ }+ }+}+```+### Choosing the Right Configuration-- **Use `core`/`lean`**: When working with basic task management workflows or when context limits are strict+- **Use `core`/`lean`**: Default setting, ideal for basic task management workflows or when context limits are strict- **Use `standard`**: For most development workflows that include task creation and analysis- **Use `all`**: When you need full functionality including tag management, dependencies, and advanced features- **Use custom list**: When you have specific tool requirements or want to experiment with minimal setsdiff --git a/apps/docs/getting-started/quick-start/configuration-quick.mdx b/apps/docs/getting-started/quick-start/configuration-quick.mdxindex 754ec2d4..0e816a8e 100644--- a/apps/docs/getting-started/quick-start/configuration-quick.mdx+++ b/apps/docs/getting-started/quick-start/configuration-quick.mdx@@ -41,9 +41,9 @@ For MCP/Cursor usage: Configure keys in the env section of your .cursor/mcp.json**Optimize Context Usage**: You can control which Task Master MCP tools are loaded using the `TASK_MASTER_TOOLS` environment variable. This helps reduce LLM context usage by only loading the tools you need.Options:-- `all` (default) - All 36 tools+- `core` or `lean` (default) - 7 essential tools- `standard` - 15 commonly used tools-- `core` or `lean` - 7 essential tools+- `all` - All 36 toolsExample:```jsondiff --git a/assets/AGENTS.md b/assets/AGENTS.mdindex a6c9f7d3..e07d72f9 100644--- a/assets/AGENTS.md+++ b/assets/AGENTS.md@@ -92,6 +92,7 @@ Task Master provides an MCP server that Claude Code can connect to. Configure in"command": "npx","args": ["-y", "task-master-ai"],"env": {+ "TASK_MASTER_TOOLS": "core","ANTHROPIC_API_KEY": "your_key_here","PERPLEXITY_API_KEY": "your_key_here","OPENAI_API_KEY": "OPENAI_API_KEY_HERE",@@ -107,6 +108,18 @@ Task Master provides an MCP server that Claude Code can connect to. Configure in}```+### MCP Tool Tiers++Default: `core` (7 tools). Set via `TASK_MASTER_TOOLS` env var.++| Tier | Count | Tools |+|------|-------|-------|+| `core` | 7 | `get_tasks`, `next_task`, `get_task`, `set_task_status`, `update_subtask`, `parse_prd`, `expand_task` |+| `standard` | 14 | core + `initialize_project`, `analyze_project_complexity`, `expand_all`, `add_subtask`, `remove_task`, `add_task`, `complexity_report` |+| `all` | 44+ | standard + dependencies, tags, research, autopilot, scoping, models, rules |++**Upgrade when tool unavailable:** Edit MCP config, change `TASK_MASTER_TOOLS` from `"core"` to `"standard"` or `"all"`, restart MCP.+### Essential MCP Tools```javascriptdiff --git a/assets/rules/taskmaster.mdc b/assets/rules/taskmaster.mdcindex 52c657e0..93ff3616 100644--- a/assets/rules/taskmaster.mdc+++ b/assets/rules/taskmaster.mdc@@ -555,4 +555,18 @@ Environment variables are used **only** for sensitive API keys related to AI pro---+## MCP Tool Tiers++Default: `core` (7 tools). Set via `TASK_MASTER_TOOLS` env var in MCP config.++| Tier | Count | Tools |+|------|-------|-------|+| `core` | 7 | `get_tasks`, `next_task`, `get_task`, `set_task_status`, `update_subtask`, `parse_prd`, `expand_task` |+| `standard` | 14 | core + `initialize_project`, `analyze_project_complexity`, `expand_all`, `add_subtask`, `remove_task`, `add_task`, `complexity_report` |+| `all` | 44+ | standard + dependencies, tags, research, autopilot, scoping, models, rules |++**Upgrade when tool unavailable:** Edit MCP config (`.cursor/mcp.json`, `.mcp.json`, or `.vscode/mcp.json`), change `TASK_MASTER_TOOLS` from `"core"` to `"standard"` or `"all"`, restart MCP.++---+For details on how these commands fit into the development process, see the [Development Workflow Guide](mdc:.cursor/rules/dev_workflow.mdc).\ No newline at end of filediff --git a/docs/configuration.md b/docs/configuration.mdindex 585ce4e6..73211042 100644--- a/docs/configuration.md+++ b/docs/configuration.md@@ -70,21 +70,20 @@ The `TASK_MASTER_TOOLS` environment variable controls which tools are loaded by#### Configuration Options-- **`all`** (default): Loads all 36 available tools (~21,000 tokens)- - Best for: Users who need the complete feature set- - Use when: Working with complex projects requiring all Task Master features- - Backward compatibility: This is the default to maintain compatibility with existing installations+- **`core`** (default, or `lean`): Loads 7 essential tools (~5,000 tokens, 70% reduction)+ - Best for: Daily development with minimal token overhead+ - Tools included: `get_tasks`, `next_task`, `get_task`, `set_task_status`, `update_subtask`, `parse_prd`, `expand_task`+ - Use when: Working in large contexts where token usage is critical+ - Note: "lean" is an alias for "core" (same tools)- **`standard`**: Loads 15 commonly used tools (~10,000 tokens, 50% reduction)- Best for: Regular task management workflows- Tools included: All core tools plus project initialization, complexity analysis, task generation, and more- Use when: You need a balanced set of features with reduced token usage-- **`core`** (or `lean`): Loads 7 essential tools (~5,000 tokens, 70% reduction)- - Best for: Daily development with minimal token overhead- - Tools included: `get_tasks`, `next_task`, `get_task`, `set_task_status`, `update_subtask`, `parse_prd`, `expand_task`- - Use when: Working in large contexts where token usage is critical- - Note: "lean" is an alias for "core" (same tools, token estimate and recommended use). You can refer to it as either "core" or "lean" when configuring.+- **`all`**: Loads all 36 available tools (~21,000 tokens)+ - Best for: Users who need the complete feature set+ - Use when: Working with complex projects requiring all Task Master features- **Custom list**: Comma-separated list of specific tool names- Best for: Specialized workflows requiring specific tools@@ -124,7 +123,7 @@ The `TASK_MASTER_TOOLS` environment variable controls which tools are loaded by#### Tool Loading Behavior-- When `TASK_MASTER_TOOLS` is unset or empty, the system defaults to `"all"`+- When `TASK_MASTER_TOOLS` is unset or empty, the system defaults to `"core"`- Invalid tool names in a user-specified list are ignored (a warning is emitted for each)- If every tool name in a custom list is invalid, the system falls back to `"all"`- Tool names are case-insensitive (e.g., `"CORE"`, `"core"`, and `"Core"` are treated identically)diff --git a/mcp-server/src/tools/index.js b/mcp-server/src/tools/index.jsindex 62682577..c1e7286e 100644--- a/mcp-server/src/tools/index.js+++ b/mcp-server/src/tools/index.js@@ -15,14 +15,14 @@ import {/*** Helper function to safely read and normalize the TASK_MASTER_TOOLS environment variable- * @returns {string} The tools configuration string, defaults to 'all'+ * @returns {string} The tools configuration string, defaults to 'core'*/export function getToolsConfiguration() {const rawValue = process.env.TASK_MASTER_TOOLS;if (!rawValue || rawValue.trim() === '') {- logger.debug('No TASK_MASTER_TOOLS env var found, defaulting to "all"');- return 'all';+ logger.debug('No TASK_MASTER_TOOLS env var found, defaulting to "core"');+ return 'core';}const normalizedValue = rawValue.trim();@@ -37,7 +37,7 @@ export function getToolsConfiguration() {* @param {string} toolMode - The tool mode configuration (defaults to 'all')* @returns {Object} Object containing registered tools, failed tools, and normalized mode*/-export function registerTaskMasterTools(server, toolMode = 'all') {+export function registerTaskMasterTools(server, toolMode = 'core') {const registeredTools = [];const failedTools = [];diff --git a/package-lock.json b/package-lock.jsonindex c264048f..0c95b9de 100644--- a/package-lock.json+++ b/package-lock.json@@ -1,12 +1,12 @@{"name": "task-master-ai",- "version": "0.37.0",+ "version": "0.37.1-rc.0","lockfileVersion": 3,"requires": true,"packages": {"": {"name": "task-master-ai",- "version": "0.37.0",+ "version": "0.37.1-rc.0","license": "MIT WITH Commons-Clause","workspaces": ["apps/*",@@ -204,7 +204,7 @@"react-dom": "^19.0.0","tailwind-merge": "^3.3.1","tailwindcss": "4.1.11",- "task-master-ai": "*",+ "task-master-ai": "0.37.1-rc.0","typescript": "^5.9.2"},"engines": {diff --git a/src/utils/create-mcp-config.js b/src/utils/create-mcp-config.jsindex 08ef9908..bd071f0c 100644--- a/src/utils/create-mcp-config.js+++ b/src/utils/create-mcp-config.js@@ -42,11 +42,15 @@ export function setupMCPConfiguration(projectRoot, mcpConfigPath) {log('info', `Setting up MCP configuration at ${mcpPath}...`);// New MCP config to be added - references the installed package+ // TASK_MASTER_TOOLS: "core" loads essential daily workflow tools (7 tools, default)+ // Options: "core" (7 essential tools), "standard" (14 common tools), "all" (44+ tools)+ // See agent rules for tool tier details and how to upgrade if neededconst newMCPServer = {'task-master-ai': {command: 'npx',args: ['-y', 'task-master-ai'],env: {+ TASK_MASTER_TOOLS: 'core',ANTHROPIC_API_KEY: 'YOUR_ANTHROPIC_API_KEY_HERE',PERPLEXITY_API_KEY: 'YOUR_PERPLEXITY_API_KEY_HERE',OPENAI_API_KEY: 'YOUR_OPENAI_KEY_HERE',@@ -158,7 +162,7 @@ export function removeTaskMasterMCPConfiguration(projectRoot, mcpConfigPath) {const mcpPath = path.join(projectRoot, mcpConfigPath);- let result = {+ const result = {success: false,removed: false,deleted: false,diff --git a/tests/unit/mcp/tools/tool-registration.test.js b/tests/unit/mcp/tools/tool-registration.test.jsindex c7fd1c73..6954a515 100644--- a/tests/unit/mcp/tools/tool-registration.test.js+++ b/tests/unit/mcp/tools/tool-registration.test.js@@ -121,20 +121,20 @@ describe('Task Master Tool Registration System', () => {});describe('Configuration Modes', () => {- it(`should register all tools (${ALL_COUNT}) when TASK_MASTER_TOOLS is not set (default behavior)`, () => {+ it(`should register core tools (${CORE_COUNT}) when toolMode defaults to 'core' (default behavior)`, () => {delete process.env.TASK_MASTER_TOOLS;registerTaskMasterTools(mockServer);expect(mockServer.addTool).toHaveBeenCalledTimes(- EXPECTED_TOOL_COUNTS.total+ EXPECTED_TOOL_COUNTS.core);});it(`should register all tools (${ALL_COUNT}) when TASK_MASTER_TOOLS=all`, () => {process.env.TASK_MASTER_TOOLS = 'all';- registerTaskMasterTools(mockServer);+ registerTaskMasterTools(mockServer, 'all');expect(mockServer.addTool).toHaveBeenCalledTimes(ALL_COUNT);});@@ -197,20 +197,20 @@ describe('Task Master Tool Registration System', () => {expect(mockServer.addTool).toHaveBeenCalledTimes(2);});- it('should default to all tools with completely invalid input', () => {+ it('should fall back to all tools with completely invalid input', () => {process.env.TASK_MASTER_TOOLS = 'completely_invalid';- registerTaskMasterTools(mockServer);+ registerTaskMasterTools(mockServer, 'completely_invalid');expect(mockServer.addTool).toHaveBeenCalledTimes(ALL_COUNT);});- it('should handle empty string environment variable', () => {+ it('should use core tools when empty string passed as toolMode', () => {process.env.TASK_MASTER_TOOLS = '';registerTaskMasterTools(mockServer);- expect(mockServer.addTool).toHaveBeenCalledTimes(ALL_COUNT);+ expect(mockServer.addTool).toHaveBeenCalledTimes(CORE_COUNT);});it('should handle whitespace in comma-separated lists', () => {@@ -232,10 +232,10 @@ describe('Task Master Tool Registration System', () => {expect(mockServer.addTool).toHaveBeenCalledTimes(2);});- it('should handle only commas and empty entries', () => {+ it('should fall back to all tools when only commas passed as toolMode', () => {process.env.TASK_MASTER_TOOLS = ',,,';- registerTaskMasterTools(mockServer);+ registerTaskMasterTools(mockServer, ',,,');expect(mockServer.addTool).toHaveBeenCalledTimes(ALL_COUNT);});@@ -254,33 +254,49 @@ describe('Task Master Tool Registration System', () => {const testCases = [{env: undefined,- expectedCount: ALL_COUNT,- description: 'undefined env (all)'+ toolMode: 'core',+ expectedCount: CORE_COUNT,+ description: 'undefined env (core default)'},{env: '',+ toolMode: 'core',+ expectedCount: CORE_COUNT,+ description: 'empty string (core default)'+ },+ {+ env: 'all',+ toolMode: 'all',expectedCount: ALL_COUNT,- description: 'empty string (all)'+ description: 'all mode'+ },+ {+ env: 'core',+ toolMode: 'core',+ expectedCount: CORE_COUNT,+ description: 'core mode'},- { env: 'all', expectedCount: ALL_COUNT, description: 'all mode' },- { env: 'core', expectedCount: CORE_COUNT, description: 'core mode' },{env: 'lean',+ toolMode: 'lean',expectedCount: CORE_COUNT,description: 'lean mode (alias)'},{env: 'standard',+ toolMode: 'standard',expectedCount: STANDARD_COUNT,description: 'standard mode'},{env: 'get_tasks,next_task',+ toolMode: 'get_tasks,next_task',expectedCount: 2,description: 'custom list'},{env: 'invalid_tool',+ toolMode: 'invalid_tool',expectedCount: ALL_COUNT,description: 'invalid fallback'}@@ -295,7 +311,7 @@ describe('Task Master Tool Registration System', () => {mockServer.tools = [];mockServer.addTool.mockClear();- registerTaskMasterTools(mockServer, testCase.env || 'all');+ registerTaskMasterTools(mockServer, testCase.toolMode);expect(mockServer.addTool).toHaveBeenCalledTimes(testCase.expectedCount@@ -308,7 +324,7 @@ describe('Task Master Tool Registration System', () => {process.env.TASK_MASTER_TOOLS = 'all';- registerTaskMasterTools(mockServer);+ registerTaskMasterTools(mockServer, 'all');const endTime = Date.now();const executionTime = endTime - startTime;@@ -401,7 +417,7 @@ describe('Task Master Tool Registration System', () => {process.env.TASK_MASTER_TOOLS = input;- expect(() => registerTaskMasterTools(mockServer)).not.toThrow();+ expect(() => registerTaskMasterTools(mockServer, input)).not.toThrow();expect(mockServer.addTool).toHaveBeenCalledTimes(ALL_COUNT);});