Files
n8n-mcp/tests/integration/mcp-protocol/tool-invocation.test.ts
Romuald Członkowski ff69e4ccca feat: Tool Consolidation - Reduce MCP Tools by 38% (v2.26.0) (#439)
* feat: Remove 9 low-value tools and consolidate n8n_health_check (v2.25.0)

Telemetry-driven tool cleanup to improve API clarity:

**Removed Tools (9):**
- list_nodes - Use search_nodes instead
- list_ai_tools - Use search_nodes with isAITool filter
- list_tasks - Low usage (0.02%)
- get_database_statistics - Use n8n_health_check
- list_templates - Use search_templates or get_templates_for_task
- get_node_as_tool_info - Documented in get_node
- validate_workflow_connections - Use validate_workflow
- validate_workflow_expressions - Use validate_workflow
- n8n_list_available_tools - Use n8n_health_check
- n8n_diagnostic - Merged into n8n_health_check

**Consolidated Tool:**
- n8n_health_check now supports mode='diagnostic' for detailed troubleshooting

**Tool Count:**
- Before: 38 tools
- After: 31 tools (18% reduction)

Concieved by Romuald Członkowski - www.aiadvisors.pl/en

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

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

* fix: cleanup stale references and update tests after tool removal

- Remove handleListAvailableTools dead code from handlers-n8n-manager.ts
- Update error messages to reference n8n_health_check(mode="diagnostic") instead of n8n_diagnostic
- Update tool counts in diagnostic messages (14 doc tools, 31 total)
- Fix error-handling.test.ts to use valid tools (search_nodes, tools_documentation)
- Remove obsolete list-tools.test.ts integration tests
- Remove unused ListToolsResponse type from response-types.ts
- Update tools.ts QUICK REFERENCE to remove list_nodes references
- Update tools-documentation.ts to remove references to removed tools
- Update tool-docs files to remove stale relatedTools references
- Fix tools.test.ts to not test removed tools (list_nodes, list_ai_tools, etc.)
- Fix parameter-validation.test.ts to not test removed tools
- Update handlers-n8n-manager.test.ts error message expectations

All 399 MCP unit tests now pass.

Conceived by Romuald Członkowski - www.aiadvisors.pl/en

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

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

* fix: update integration tests to use valid tools after v2.25.0 removal

Replaced all references to removed tools in integration tests:
- list_nodes -> search_nodes
- get_database_statistics -> tools_documentation
- list_ai_tools -> search_nodes/tools_documentation
- list_tasks -> tools_documentation
- get_node_as_tool_info -> removed test section

Updated test files:
- tests/integration/mcp-protocol/basic-connection.test.ts
- tests/integration/mcp-protocol/performance.test.ts
- tests/integration/mcp-protocol/session-management.test.ts
- tests/integration/mcp-protocol/test-helpers.ts
- tests/integration/mcp-protocol/tool-invocation.test.ts
- tests/integration/telemetry/mcp-telemetry.test.ts
- tests/unit/mcp/disabled-tools.test.ts
- tests/unit/mcp/tools-documentation.test.ts

Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en

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

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

* feat: Tool consolidation v2.26.0 - reduce tools by 38% (31 → 19)

Major consolidation of MCP tools using mode-based parameters for better
AI agent ergonomics:

Node Tools:
- get_node_documentation → get_node with mode='documentation'
- search_node_properties → get_node with mode='search_properties'
- get_property_dependencies → removed

Validation Tools:
- validate_node_operation + validate_node_minimal → validate_node with mode param

Template Tools:
- list_node_templates → search_templates with searchMode='nodes'
- search_templates_by_metadata → search_templates with searchMode='metadata'
- get_templates_for_task → search_templates with searchMode='task'

Workflow Getters:
- n8n_get_workflow_details/structure/minimal → n8n_get_workflow with mode param

Execution Tools:
- n8n_list/get/delete_execution → n8n_executions with action param

Test updates for all consolidated tools.

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

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

Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en

* docs: comprehensive README update for v2.26.0 tool consolidation

- Quick Start: Added hosted service (dashboard.n8n-mcp.com) as primary option
- Self-hosting: Renamed options to A (npx), B (Docker), C (Local), D (Railway)
- Removed: "Memory Leak Fix (v2.20.2)" section (outdated)
- Removed: "Known Issues" section (outdated container management)
- Claude Project Setup: Updated all tool references to v2.26.0 consolidated tools
  - validate_node({mode: 'minimal'|'full'}) instead of separate tools
  - search_templates({searchMode: ...}) unified template search
  - get_node({mode: 'docs'|'search_properties'}) for documentation
  - n8n_executions({action: ...}) unified execution management
- Available MCP Tools: Updated to show 19 consolidated tools (7 core + 12 mgmt)
- Recent Updates: Simplified to just link to CHANGELOG.md

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

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

Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en

* fix: update tool count from 31 to 19 in diagnostic message

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

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

* fix(tests): update tool count expectations for v2.26.0

Update handlers-n8n-manager.test.ts to expect new consolidated
tool counts (7/12/19) after v2.26.0 tool consolidation.

Conceived by Romuald Członkowski - www.aiadvisors.pl/en

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

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

---------

Co-authored-by: Claude <noreply@anthropic.com>
2025-11-25 18:39:00 +01:00

438 lines
15 KiB
TypeScript

import { describe, it, expect, beforeEach, afterEach } from 'vitest';
import { InMemoryTransport } from '@modelcontextprotocol/sdk/inMemory.js';
import { Client } from '@modelcontextprotocol/sdk/client/index.js';
import { TestableN8NMCPServer } from './test-helpers';
describe('MCP Tool Invocation', () => {
let mcpServer: TestableN8NMCPServer;
let client: Client;
beforeEach(async () => {
mcpServer = new TestableN8NMCPServer();
await mcpServer.initialize();
const [serverTransport, clientTransport] = InMemoryTransport.createLinkedPair();
await mcpServer.connectToTransport(serverTransport);
client = new Client({
name: 'test-client',
version: '1.0.0'
}, {
capabilities: {}
});
await client.connect(clientTransport);
});
afterEach(async () => {
await client.close();
await mcpServer.close();
});
describe('Node Discovery Tools', () => {
describe('search_nodes', () => {
it('should search nodes by keyword', async () => {
const response = await client.callTool({ name: 'search_nodes', arguments: {
query: 'webhook'
}});
const result = JSON.parse(((response as any).content[0]).text);
const nodes = result.results;
expect(nodes.length).toBeGreaterThan(0);
// Should find webhook node
const webhookNode = nodes.find((n: any) => n.displayName.toLowerCase().includes('webhook'));
expect(webhookNode).toBeDefined();
});
it('should support different search modes', async () => {
// OR mode
const orResponse = await client.callTool({ name: 'search_nodes', arguments: {
query: 'http request',
mode: 'OR'
}});
const orResult = JSON.parse(((orResponse as any).content[0]).text);
const orNodes = orResult.results;
expect(orNodes.length).toBeGreaterThan(0);
// AND mode
const andResponse = await client.callTool({ name: 'search_nodes', arguments: {
query: 'http request',
mode: 'AND'
}});
const andResult = JSON.parse(((andResponse as any).content[0]).text);
const andNodes = andResult.results;
expect(andNodes.length).toBeLessThanOrEqual(orNodes.length);
// FUZZY mode - use less typo-heavy search
const fuzzyResponse = await client.callTool({ name: 'search_nodes', arguments: {
query: 'http req', // Partial match should work
mode: 'FUZZY'
}});
const fuzzyResult = JSON.parse(((fuzzyResponse as any).content[0]).text);
const fuzzyNodes = fuzzyResult.results;
expect(fuzzyNodes.length).toBeGreaterThan(0);
});
it('should respect result limit', async () => {
const response = await client.callTool({ name: 'search_nodes', arguments: {
query: 'node',
limit: 3
}});
const result = JSON.parse(((response as any).content[0]).text);
const nodes = result.results;
expect(nodes).toHaveLength(3);
});
});
describe('get_node', () => {
it('should get complete node information', async () => {
const response = await client.callTool({ name: 'get_node', arguments: {
nodeType: 'nodes-base.httpRequest',
detail: 'full'
}});
expect(((response as any).content[0]).type).toBe('text');
const nodeInfo = JSON.parse(((response as any).content[0]).text);
expect(nodeInfo).toHaveProperty('nodeType', 'nodes-base.httpRequest');
expect(nodeInfo).toHaveProperty('displayName');
expect(nodeInfo).toHaveProperty('description');
expect(nodeInfo).toHaveProperty('version');
});
it('should handle non-existent nodes', async () => {
try {
await client.callTool({ name: 'get_node', arguments: {
nodeType: 'nodes-base.nonExistent'
}});
expect.fail('Should have thrown an error');
} catch (error: any) {
expect(error.message).toContain('not found');
}
});
it('should handle invalid node type format', async () => {
try {
await client.callTool({ name: 'get_node', arguments: {
nodeType: 'invalidFormat'
}});
expect.fail('Should have thrown an error');
} catch (error: any) {
expect(error.message).toContain('not found');
}
});
});
describe('get_node with different detail levels', () => {
it('should return standard detail by default', async () => {
const response = await client.callTool({ name: 'get_node', arguments: {
nodeType: 'nodes-base.httpRequest'
}});
const nodeInfo = JSON.parse(((response as any).content[0]).text);
expect(nodeInfo).toHaveProperty('nodeType');
expect(nodeInfo).toHaveProperty('displayName');
expect(nodeInfo).toHaveProperty('description');
expect(nodeInfo).toHaveProperty('requiredProperties');
expect(nodeInfo).toHaveProperty('commonProperties');
// Should be smaller than full detail
const fullResponse = await client.callTool({ name: 'get_node', arguments: {
nodeType: 'nodes-base.httpRequest',
detail: 'full'
}});
expect(((response as any).content[0]).text.length).toBeLessThan(((fullResponse as any).content[0]).text.length);
});
});
});
describe('Validation Tools', () => {
// v2.26.0: validate_node_operation consolidated into validate_node with mode parameter
describe('validate_node', () => {
it('should validate valid node configuration', async () => {
const response = await client.callTool({ name: 'validate_node', arguments: {
nodeType: 'nodes-base.httpRequest',
config: {
method: 'GET',
url: 'https://api.example.com/data'
},
mode: 'full'
}});
const validation = JSON.parse(((response as any).content[0]).text);
expect(validation).toHaveProperty('valid');
expect(validation).toHaveProperty('errors');
expect(validation).toHaveProperty('warnings');
});
it('should detect missing required fields', async () => {
const response = await client.callTool({ name: 'validate_node', arguments: {
nodeType: 'nodes-base.httpRequest',
config: {
method: 'GET'
// Missing required 'url' field
},
mode: 'full'
}});
const validation = JSON.parse(((response as any).content[0]).text);
expect(validation.valid).toBe(false);
expect(validation.errors.length).toBeGreaterThan(0);
expect(validation.errors[0].message.toLowerCase()).toContain('url');
});
it('should support different validation profiles', async () => {
const profiles = ['minimal', 'runtime', 'ai-friendly', 'strict'];
for (const profile of profiles) {
const response = await client.callTool({ name: 'validate_node', arguments: {
nodeType: 'nodes-base.httpRequest',
config: { method: 'GET', url: 'https://api.example.com' },
mode: 'full',
profile
}});
const validation = JSON.parse(((response as any).content[0]).text);
expect(validation).toHaveProperty('profile', profile);
}
});
});
describe('validate_workflow', () => {
it('should validate complete workflow', async () => {
const workflow = {
nodes: [
{
id: '1',
name: 'Start',
type: 'nodes-base.manualTrigger',
typeVersion: 1,
position: [0, 0],
parameters: {}
},
{
id: '2',
name: 'HTTP Request',
type: 'nodes-base.httpRequest',
typeVersion: 3,
position: [250, 0],
parameters: {
method: 'GET',
url: 'https://api.example.com/data'
}
}
],
connections: {
'Start': {
'main': [[{ node: 'HTTP Request', type: 'main', index: 0 }]]
}
}
};
const response = await client.callTool({ name: 'validate_workflow', arguments: {
workflow
}});
const validation = JSON.parse(((response as any).content[0]).text);
expect(validation).toHaveProperty('valid');
expect(validation).toHaveProperty('errors');
expect(validation).toHaveProperty('warnings');
});
it('should detect connection errors', async () => {
const workflow = {
nodes: [
{
id: '1',
name: 'Start',
type: 'nodes-base.manualTrigger',
typeVersion: 1,
position: [0, 0],
parameters: {}
}
],
connections: {
'Start': {
'main': [[{ node: 'NonExistent', type: 'main', index: 0 }]]
}
}
};
const response = await client.callTool({ name: 'validate_workflow', arguments: {
workflow
}});
const validation = JSON.parse(((response as any).content[0]).text);
expect(validation.valid).toBe(false);
expect(validation.errors.length).toBeGreaterThan(0);
});
it('should validate expressions', async () => {
const workflow = {
nodes: [
{
id: '1',
name: 'Start',
type: 'n8n-nodes-base.manualTrigger',
typeVersion: 1,
position: [0, 0],
parameters: {}
},
{
id: '2',
name: 'Set',
type: 'n8n-nodes-base.set',
typeVersion: 3.4,
position: [250, 0],
parameters: {
mode: 'manual',
duplicateItem: false,
values: {
string: [
{
name: 'test',
value: '={{ $json.invalidExpression }}'
}
]
}
}
}
],
connections: {
'Start': {
'main': [[{ node: 'Set', type: 'main', index: 0 }]]
}
}
};
const response = await client.callTool({ name: 'validate_workflow', arguments: {
workflow,
options: {
validateExpressions: true
}
}});
const validation = JSON.parse(((response as any).content[0]).text);
expect(validation).toHaveProperty('valid');
// The workflow should have either errors or warnings about the expression
if (validation.errors && validation.errors.length > 0) {
expect(validation.errors.some((e: any) =>
e.message.includes('expression') || e.message.includes('$json')
)).toBe(true);
} else if (validation.warnings) {
expect(validation.warnings.length).toBeGreaterThan(0);
expect(validation.warnings.some((w: any) =>
w.message.includes('expression') || w.message.includes('$json')
)).toBe(true);
}
});
});
});
describe('Documentation Tools', () => {
describe('tools_documentation', () => {
it('should get quick start guide', async () => {
const response = await client.callTool({ name: 'tools_documentation', arguments: {} });
expect(((response as any).content[0]).type).toBe('text');
expect(((response as any).content[0]).text).toContain('n8n MCP Tools');
});
it('should get specific tool documentation', async () => {
const response = await client.callTool({ name: 'tools_documentation', arguments: {
topic: 'search_nodes'
}});
expect(((response as any).content[0]).text).toContain('search_nodes');
expect(((response as any).content[0]).text).toContain('Text search');
});
it('should get comprehensive documentation', async () => {
const response = await client.callTool({ name: 'tools_documentation', arguments: {
depth: 'full'
}});
expect(((response as any).content[0]).text.length).toBeGreaterThan(5000);
expect(((response as any).content[0]).text).toBeDefined();
});
it('should handle invalid topics gracefully', async () => {
const response = await client.callTool({ name: 'tools_documentation', arguments: {
topic: 'nonexistent_tool'
}});
expect(((response as any).content[0]).text).toContain('not found');
});
});
});
// AI Tools section removed - list_ai_tools and get_node_as_tool_info were removed in v2.25.0
// Use search_nodes with query for finding AI-capable nodes
describe('Complex Tool Interactions', () => {
it('should handle tool chaining', async () => {
// Search for nodes
const searchResponse = await client.callTool({ name: 'search_nodes', arguments: {
query: 'slack'
}});
const searchResult = JSON.parse(((searchResponse as any).content[0]).text);
const nodes = searchResult.results;
// Get info for first result
const firstNode = nodes[0];
const infoResponse = await client.callTool({ name: 'get_node', arguments: {
nodeType: firstNode.nodeType
}});
expect(((infoResponse as any).content[0]).text).toContain(firstNode.displayName);
});
it('should handle parallel tool calls', async () => {
const toolCalls = [
{ name: 'search_nodes', arguments: { query: 'http' } },
{ name: 'tools_documentation', arguments: {} },
{ name: 'get_node', arguments: { nodeType: 'nodes-base.httpRequest' } },
{ name: 'search_nodes', arguments: { query: 'webhook' } }
];
const promises = toolCalls.map(call =>
client.callTool(call)
);
const responses = await Promise.all(promises);
expect(responses).toHaveLength(toolCalls.length);
responses.forEach(response => {
expect(response.content).toHaveLength(1);
expect(((response as any).content[0]).type).toBe('text');
});
});
it('should maintain consistency across related tools', async () => {
// Get node via different methods
const nodeType = 'nodes-base.httpRequest';
const [fullInfo, essentials, searchResult] = await Promise.all([
client.callTool({ name: 'get_node', arguments: { nodeType } }),
client.callTool({ name: 'get_node', arguments: { nodeType } }),
client.callTool({ name: 'search_nodes', arguments: { query: 'httpRequest' } })
]);
const full = JSON.parse(((fullInfo as any).content[0]).text);
const essential = JSON.parse(((essentials as any).content[0]).text);
const searchData = JSON.parse(((searchResult as any).content[0]).text);
const search = searchData.results;
// Should all reference the same node
expect(full.nodeType).toBe('nodes-base.httpRequest');
expect(essential.displayName).toBe(full.displayName);
expect(search.find((n: any) => n.nodeType === 'nodes-base.httpRequest')).toBeDefined();
});
});
});