@runhuman/mcp-server 1.1.0 → 2.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (4) hide show
  1. package/.env.example +12 -12
  2. package/README.md +219 -219
  3. package/dist/index.js +83 -83
  4. package/package.json +62 -62
package/.env.example CHANGED
@@ -1,12 +1,12 @@
1
- # RunHuman API Configuration
2
-
3
- # API Base URL
4
- # For local development, use http://localhost:3400
5
- # For production, use your deployed API URL
6
- RUNHUMAN_API_URL=http://localhost:3400
7
-
8
- # API Key
9
- # Get this from the API dashboard at http://localhost:3400/app.html
10
- # Or use the default test key below for local development
11
- # Format: qa_live_xxxxxxxxxxxxxxxxxxxxx
12
- RUNHUMAN_API_KEY=qa_live_test_key_for_demo_purposes_only_12345
1
+ # RunHuman API Configuration
2
+
3
+ # API Base URL
4
+ # For local development, use http://localhost:3400
5
+ # For production, use your deployed API URL
6
+ RUNHUMAN_API_URL=http://localhost:3400
7
+
8
+ # API Key
9
+ # Get this from the API dashboard at http://localhost:3400/app.html
10
+ # Or use the default test key below for local development
11
+ # Format: qa_live_xxxxxxxxxxxxxxxxxxxxx
12
+ RUNHUMAN_API_KEY=qa_live_test_key_for_demo_purposes_only_12345
package/README.md CHANGED
@@ -1,219 +1,219 @@
1
- # RunHuman MCP Server
2
-
3
- A Model Context Protocol (MCP) server that allows AI agents to interact with the RunHuman QA testing service.
4
-
5
- ## Overview
6
-
7
- This MCP server provides tools for creating and managing human QA jobs through the RunHuman API. AI agents can use this server to:
8
-
9
- - Create new QA jobs with custom schemas
10
- - Check the status of running jobs
11
- - Retrieve completed job results
12
-
13
- ## Installation
14
-
15
- ### For Claude Desktop (Recommended)
16
-
17
- 1. Get your API key at: https://runhuman.com/app.html
18
-
19
- 2. Add to your Claude Desktop config (`~/Library/Application Support/Claude/claude_desktop_config.json` on Mac):
20
-
21
- ```json
22
- {
23
- "mcpServers": {
24
- "runhuman": {
25
- "command": "npx",
26
- "args": ["-y", "@runhuman/mcp-server", "--api-key=qa_live_xxxxxxxxxxxxx"]
27
- }
28
- }
29
- }
30
- ```
31
-
32
- 3. Restart Claude Desktop
33
-
34
- That's it! The server will be automatically downloaded and run by Claude.
35
-
36
- ### For Development
37
-
38
- From the monorepo root:
39
-
40
- ```bash
41
- npm install
42
- npm run build --workspace=@runhuman/mcp-server
43
-
44
- # Run with API key
45
- node packages/mcp-server/dist/index.js --api-key=qa_live_xxxxx
46
- ```
47
-
48
- ### Available Tools
49
-
50
- #### `create_job`
51
- Create a new QA job with human testers.
52
-
53
- **Parameters:**
54
- - `url` (string): The URL to test
55
- - `description` (string): Instructions for the human tester describing what to test
56
- - `schema` (object): Expected result schema that the tester response will be extracted into
57
-
58
- #### `get_job_status`
59
- Get the current status of a QA job.
60
-
61
- **Parameters:**
62
- - `jobId` (string): The ID of the job to check
63
-
64
- #### `get_job_result`
65
- Get the results of a completed QA job.
66
-
67
- **Parameters:**
68
- - `jobId` (string): The ID of the completed job
69
-
70
- **Returns:**
71
- - `result`: Structured test results extracted from tester's response
72
- - `status`: Job status (completed, failed, etc.)
73
- - `costUsd`: Exact cost in USD with full precision (e.g., 0.396)
74
- - `testDurationSeconds`: Time spent by tester in seconds (rounded up)
75
- - Additional metadata (timestamps, tester info, etc.)
76
-
77
- **Cost Calculation:**
78
- - Costs are calculated as: `duration × $0.0018/second` (general-use tier)
79
- - Duration is always rounded UP using Math.ceil (any partial second counts)
80
- - Costs are never $0 unless the tester never actually worked on it
81
- - Full precision maintained (not rounded to cents)
82
-
83
- ## Configuration
84
-
85
- The MCP server needs to be configured with your RunHuman API credentials.
86
-
87
- ### 1. Get an API Key
88
-
89
- **Option A: Via Dashboard**
90
- 1. Start the API server: `npm run dev --workspace=@runhuman/api`
91
- 2. Open http://localhost:3400/app.html
92
- 3. Go to "API Keys" tab
93
- 4. Click "Create API Key"
94
- 5. Copy the key (starts with `qa_live_`)
95
-
96
- **Option B: Use Default Test Key**
97
- - For local development, you can use: `qa_live_test_key_123`
98
- - This key exists in `packages/api/data/api-keys.json`
99
-
100
- ### 2. Configure Environment Variables
101
-
102
- Create a `.env` file in the MCP server directory:
103
-
104
- ```bash
105
- # For local development
106
- RUNHUMAN_API_URL=http://localhost:3400
107
- RUNHUMAN_API_KEY=qa_live_test_key_123
108
-
109
- # For production
110
- RUNHUMAN_API_URL=https://api.runhuman.com
111
- RUNHUMAN_API_KEY=qa_live_xxxxxxxxxxxxxxxxxxxxx
112
- ```
113
-
114
- **Important:** Never commit `.env` files to git! They're already in `.gitignore`.
115
-
116
- ### 3. Verify Configuration
117
-
118
- Test your API key works:
119
-
120
- ```bash
121
- curl http://localhost:3400/api/jobs \
122
- -H "Authorization: Bearer qa_live_test_key_123" \
123
- -H "Content-Type: application/json" \
124
- -d '{"url":"https://example.com","description":"test","outputSchema":{}}'
125
- ```
126
-
127
- Should return a job ID if authentication works.
128
-
129
- For more details, see [docs/API-AUTHENTICATION.md](docs/API-AUTHENTICATION.md)
130
-
131
- ## Testing
132
-
133
- The MCP server includes automated tests to verify it's working correctly:
134
-
135
- ```bash
136
- # Build first
137
- npm run build --workspace=@runhuman/mcp-server
138
-
139
- # Run simple automated test
140
- npm run test --workspace=@runhuman/mcp-server
141
-
142
- # Or use the MCP Inspector (interactive testing)
143
- npm run test:inspector --workspace=@runhuman/mcp-server
144
- ```
145
-
146
- The test script will:
147
- 1. ✅ Initialize a connection to the MCP server
148
- 2. ✅ List all available tools (create_job, get_job_status, get_job_result)
149
- 3. ✅ Test calling the create_job tool
150
-
151
- ### Expected Test Output
152
-
153
- ```
154
- ✅ Server initialized successfully
155
- ✅ Tools listed: create_job, get_job_status, get_job_result
156
- ✅ create_job tool called successfully
157
- ```
158
-
159
- ## Development
160
-
161
- ```bash
162
- # Watch mode (auto-rebuild on changes)
163
- npm run dev --workspace=@runhuman/mcp-server
164
-
165
- # Build
166
- npm run build --workspace=@runhuman/mcp-server
167
-
168
- # Test after building
169
- npm run test --workspace=@runhuman/mcp-server
170
- ```
171
-
172
- ## Integration with Claude Desktop
173
-
174
- To use this MCP server with Claude Desktop, add it to your configuration:
175
-
176
- ```json
177
- {
178
- "mcpServers": {
179
- "runhuman": {
180
- "command": "node",
181
- "args": ["/path/to/qa-experiment/packages/mcp-server/dist/index.js"]
182
- }
183
- }
184
- }
185
- ```
186
-
187
- ## Example Usage
188
-
189
- Once connected to an AI agent (like Claude), the agent can use these tools naturally:
190
-
191
- **User:** "Can someone test my checkout page at https://myapp.com/checkout?"
192
-
193
- **Agent uses create_job:**
194
- ```
195
- ✅ Job created successfully!
196
- Job ID: job_abc123
197
- Status: pending
198
- ...
199
- ```
200
-
201
- **Agent polls get_job_status until complete, then calls get_job_result:**
202
- ```
203
- ✅ Test completed!
204
- Results Summary:
205
- - Checkout Flow: ✅ Working
206
- - Payment Processing: ✅ Successful
207
- ...
208
- ```
209
-
210
- ## Developer Documentation
211
-
212
- For developers working on this MCP server:
213
- - [docs/HOW-AGENTS-USE-MCP.md](docs/HOW-AGENTS-USE-MCP.md) - How AI agents discover and use MCP servers
214
- - [docs/TOOL-RESPONSE-BEST-PRACTICES.md](docs/TOOL-RESPONSE-BEST-PRACTICES.md) - Best practices for tool responses
215
-
216
- ## Learn More
217
-
218
- - [Model Context Protocol Documentation](https://modelcontextprotocol.io/)
219
- - [RunHuman API Documentation](../api/README.md)
1
+ # RunHuman MCP Server
2
+
3
+ A Model Context Protocol (MCP) server that allows AI agents to interact with the RunHuman QA testing service.
4
+
5
+ ## Overview
6
+
7
+ This MCP server provides tools for creating and managing human QA jobs through the RunHuman API. AI agents can use this server to:
8
+
9
+ - Create new QA jobs with custom schemas
10
+ - Check the status of running jobs
11
+ - Retrieve completed job results
12
+
13
+ ## Installation
14
+
15
+ ### For Claude Desktop (Recommended)
16
+
17
+ 1. Get your API key at: https://runhuman.com/app.html
18
+
19
+ 2. Add to your Claude Desktop config (`~/Library/Application Support/Claude/claude_desktop_config.json` on Mac):
20
+
21
+ ```json
22
+ {
23
+ "mcpServers": {
24
+ "runhuman": {
25
+ "command": "npx",
26
+ "args": ["-y", "@runhuman/mcp-server", "--api-key=qa_live_xxxxxxxxxxxxx"]
27
+ }
28
+ }
29
+ }
30
+ ```
31
+
32
+ 3. Restart Claude Desktop
33
+
34
+ That's it! The server will be automatically downloaded and run by Claude.
35
+
36
+ ### For Development
37
+
38
+ From the monorepo root:
39
+
40
+ ```bash
41
+ npm install
42
+ npm run build --workspace=@runhuman/mcp-server
43
+
44
+ # Run with API key
45
+ node packages/mcp-server/dist/index.js --api-key=qa_live_xxxxx
46
+ ```
47
+
48
+ ### Available Tools
49
+
50
+ #### `create_job`
51
+ Create a new QA job with human testers.
52
+
53
+ **Parameters:**
54
+ - `url` (string): The URL to test
55
+ - `description` (string): Instructions for the human tester describing what to test
56
+ - `schema` (object): Expected result schema that the tester response will be extracted into
57
+
58
+ #### `get_job_status`
59
+ Get the current status of a QA job.
60
+
61
+ **Parameters:**
62
+ - `jobId` (string): The ID of the job to check
63
+
64
+ #### `get_job_result`
65
+ Get the results of a completed QA job.
66
+
67
+ **Parameters:**
68
+ - `jobId` (string): The ID of the completed job
69
+
70
+ **Returns:**
71
+ - `result`: Structured test results extracted from tester's response
72
+ - `status`: Job status (completed, failed, etc.)
73
+ - `costUsd`: Exact cost in USD with full precision (e.g., 0.396)
74
+ - `testDurationSeconds`: Time spent by tester in seconds (rounded up)
75
+ - Additional metadata (timestamps, tester info, etc.)
76
+
77
+ **Cost Calculation:**
78
+ - Costs are calculated as: `duration × $0.0018/second` (general-use tier)
79
+ - Duration is always rounded UP using Math.ceil (any partial second counts)
80
+ - Costs are never $0 unless the tester never actually worked on it
81
+ - Full precision maintained (not rounded to cents)
82
+
83
+ ## Configuration
84
+
85
+ The MCP server needs to be configured with your RunHuman API credentials.
86
+
87
+ ### 1. Get an API Key
88
+
89
+ **Option A: Via Dashboard**
90
+ 1. Start the API server: `npm run dev --workspace=@runhuman/api`
91
+ 2. Open http://localhost:3400/app.html
92
+ 3. Go to "API Keys" tab
93
+ 4. Click "Create API Key"
94
+ 5. Copy the key (starts with `qa_live_`)
95
+
96
+ **Option B: Use Default Test Key**
97
+ - For local development, you can use: `qa_live_test_key_123`
98
+ - This key exists in `packages/api/data/api-keys.json`
99
+
100
+ ### 2. Configure Environment Variables
101
+
102
+ Create a `.env` file in the MCP server directory:
103
+
104
+ ```bash
105
+ # For local development
106
+ RUNHUMAN_API_URL=http://localhost:3400
107
+ RUNHUMAN_API_KEY=qa_live_test_key_123
108
+
109
+ # For production
110
+ RUNHUMAN_API_URL=https://api.runhuman.com
111
+ RUNHUMAN_API_KEY=qa_live_xxxxxxxxxxxxxxxxxxxxx
112
+ ```
113
+
114
+ **Important:** Never commit `.env` files to git! They're already in `.gitignore`.
115
+
116
+ ### 3. Verify Configuration
117
+
118
+ Test your API key works:
119
+
120
+ ```bash
121
+ curl http://localhost:3400/api/jobs \
122
+ -H "Authorization: Bearer qa_live_test_key_123" \
123
+ -H "Content-Type: application/json" \
124
+ -d '{"url":"https://example.com","description":"test","outputSchema":{}}'
125
+ ```
126
+
127
+ Should return a job ID if authentication works.
128
+
129
+ For more details, see [docs/API-AUTHENTICATION.md](docs/API-AUTHENTICATION.md)
130
+
131
+ ## Testing
132
+
133
+ The MCP server includes automated tests to verify it's working correctly:
134
+
135
+ ```bash
136
+ # Build first
137
+ npm run build --workspace=@runhuman/mcp-server
138
+
139
+ # Run simple automated test
140
+ npm run test --workspace=@runhuman/mcp-server
141
+
142
+ # Or use the MCP Inspector (interactive testing)
143
+ npm run test:inspector --workspace=@runhuman/mcp-server
144
+ ```
145
+
146
+ The test script will:
147
+ 1. ✅ Initialize a connection to the MCP server
148
+ 2. ✅ List all available tools (create_job, get_job_status, get_job_result)
149
+ 3. ✅ Test calling the create_job tool
150
+
151
+ ### Expected Test Output
152
+
153
+ ```
154
+ ✅ Server initialized successfully
155
+ ✅ Tools listed: create_job, get_job_status, get_job_result
156
+ ✅ create_job tool called successfully
157
+ ```
158
+
159
+ ## Development
160
+
161
+ ```bash
162
+ # Watch mode (auto-rebuild on changes)
163
+ npm run dev --workspace=@runhuman/mcp-server
164
+
165
+ # Build
166
+ npm run build --workspace=@runhuman/mcp-server
167
+
168
+ # Test after building
169
+ npm run test --workspace=@runhuman/mcp-server
170
+ ```
171
+
172
+ ## Integration with Claude Desktop
173
+
174
+ To use this MCP server with Claude Desktop, add it to your configuration:
175
+
176
+ ```json
177
+ {
178
+ "mcpServers": {
179
+ "runhuman": {
180
+ "command": "node",
181
+ "args": ["/path/to/qa-experiment/packages/mcp-server/dist/index.js"]
182
+ }
183
+ }
184
+ }
185
+ ```
186
+
187
+ ## Example Usage
188
+
189
+ Once connected to an AI agent (like Claude), the agent can use these tools naturally:
190
+
191
+ **User:** "Can someone test my checkout page at https://myapp.com/checkout?"
192
+
193
+ **Agent uses create_job:**
194
+ ```
195
+ ✅ Job created successfully!
196
+ Job ID: job_abc123
197
+ Status: pending
198
+ ...
199
+ ```
200
+
201
+ **Agent polls get_job_status until complete, then calls get_job_result:**
202
+ ```
203
+ ✅ Test completed!
204
+ Results Summary:
205
+ - Checkout Flow: ✅ Working
206
+ - Payment Processing: ✅ Successful
207
+ ...
208
+ ```
209
+
210
+ ## Developer Documentation
211
+
212
+ For developers working on this MCP server:
213
+ - [docs/HOW-AGENTS-USE-MCP.md](docs/HOW-AGENTS-USE-MCP.md) - How AI agents discover and use MCP servers
214
+ - [docs/TOOL-RESPONSE-BEST-PRACTICES.md](docs/TOOL-RESPONSE-BEST-PRACTICES.md) - Best practices for tool responses
215
+
216
+ ## Learn More
217
+
218
+ - [Model Context Protocol Documentation](https://modelcontextprotocol.io/)
219
+ - [RunHuman API Documentation](../api/README.md)
package/dist/index.js CHANGED
@@ -75,22 +75,22 @@ server.setRequestHandler(ListToolsRequestSchema, async () => {
75
75
  tools: [
76
76
  {
77
77
  name: 'create_job',
78
- description: `⚠️ IMPORTANT: This ONLY creates and queues a job. It does NOT perform the test or return results. You MUST follow up with get_job_status and get_job_result.
79
-
80
- Creates a QA job that will be performed by a REAL HUMAN tester (not AI). The human will manually test your application, describe findings in natural language, and GPT-4o will extract structured data from their response.
81
-
82
- Use this when you need human verification of:
83
- - UI/UX functionality that's hard to automate
84
- - Visual issues, accessibility problems
85
- - Complex user flows (login, checkout, forms)
86
- - Cross-browser compatibility
87
- - Real user experience feedback
88
-
89
- ⚠️ REQUIRED WORKFLOW (do NOT skip steps):
90
- 1. create_job → Returns jobId (job is now QUEUED, not complete!)
91
- 2. get_job_status → Poll every 30-60 seconds until status="completed" (takes 2-10 min)
92
- 3. get_job_result → Retrieve the actual test results
93
-
78
+ description: `⚠️ IMPORTANT: This ONLY creates and queues a job. It does NOT perform the test or return results. You MUST follow up with get_job_status and get_job_result.
79
+
80
+ Creates a QA job that will be performed by a REAL HUMAN tester (not AI). The human will manually test your application, describe findings in natural language, and GPT-4o will extract structured data from their response.
81
+
82
+ Use this when you need human verification of:
83
+ - UI/UX functionality that's hard to automate
84
+ - Visual issues, accessibility problems
85
+ - Complex user flows (login, checkout, forms)
86
+ - Cross-browser compatibility
87
+ - Real user experience feedback
88
+
89
+ ⚠️ REQUIRED WORKFLOW (do NOT skip steps):
90
+ 1. create_job → Returns jobId (job is now QUEUED, not complete!)
91
+ 2. get_job_status → Poll every 30-60 seconds until status="completed" (takes 2-10 min)
92
+ 3. get_job_result → Retrieve the actual test results
93
+
94
94
  DO NOT treat job creation as completion. You MUST wait for and retrieve results.`,
95
95
  inputSchema: {
96
96
  type: 'object',
@@ -117,10 +117,10 @@ DO NOT treat job creation as completion. You MUST wait for and retrieve results.
117
117
  },
118
118
  {
119
119
  name: 'get_job_status',
120
- description: `Check the current status of a QA job. Jobs progress through states: pending → claimed → in_progress → completed (or failed/timeout).
121
-
122
- Use this to poll for completion before fetching results. Typical job completion time is 2-10 minutes depending on test complexity.
123
-
120
+ description: `Check the current status of a QA job. Jobs progress through states: pending → claimed → in_progress → completed (or failed/timeout).
121
+
122
+ Use this to poll for completion before fetching results. Typical job completion time is 2-10 minutes depending on test complexity.
123
+
124
124
  Returns: { status: "pending" | "claimed" | "in_progress" | "completed" | "failed" | "timeout", message: "..." }`,
125
125
  inputSchema: {
126
126
  type: 'object',
@@ -135,10 +135,10 @@ Returns: { status: "pending" | "claimed" | "in_progress" | "completed" | "failed
135
135
  },
136
136
  {
137
137
  name: 'get_job_result',
138
- description: `Get the structured results of a completed QA job. Only call this after get_job_status shows status="completed".
139
-
140
- Returns the tester's response extracted into your specified schema, plus metadata about timing and the raw tester response.
141
-
138
+ description: `Get the structured results of a completed QA job. Only call this after get_job_status shows status="completed".
139
+
140
+ Returns the tester's response extracted into your specified schema, plus metadata about timing and the raw tester response.
141
+
142
142
  If the job isn't complete yet, returns an error. If extraction failed, includes the raw response so you can see what the tester said.`,
143
143
  inputSchema: {
144
144
  type: 'object',
@@ -184,14 +184,14 @@ server.setRequestHandler(CallToolRequestSchema, async (request) => {
184
184
  return {
185
185
  content: [{
186
186
  type: 'text',
187
- text: `❌ Failed to create job
188
-
189
- Error: ${error.error || error.message || response.statusText}
190
- Status: ${response.status}
191
-
192
- Please check:
193
- - Your RUNHUMAN_API_KEY is valid
194
- - The API server is running at ${API_URL}
187
+ text: `❌ Failed to create job
188
+
189
+ Error: ${error.error || error.message || response.statusText}
190
+ Status: ${response.status}
191
+
192
+ Please check:
193
+ - Your RUNHUMAN_API_KEY is valid
194
+ - The API server is running at ${API_URL}
195
195
  - Your API key has permission to create jobs`
196
196
  }],
197
197
  isError: true
@@ -203,21 +203,21 @@ Please check:
203
203
  content: [
204
204
  {
205
205
  type: 'text',
206
- text: `⏳ Job QUEUED (not complete yet!)
207
-
208
- Job ID: ${data.jobId}
209
- Status: ${data.status}
210
- URL: ${args.url}
211
- Time limit: ${duration} minutes
212
-
213
- The test has been queued for a HUMAN tester. It is NOT complete yet.
214
-
215
- ⚠️ REQUIRED NEXT STEPS:
216
- 1. Wait 30-60 seconds
217
- 2. Call get_job_status(jobId: "${data.jobId}") to check if complete
218
- 3. Repeat step 2 until status="completed" (typical time: 2-10 minutes)
219
- 4. Call get_job_result(jobId: "${data.jobId}") to get the actual test results
220
-
206
+ text: `⏳ Job QUEUED (not complete yet!)
207
+
208
+ Job ID: ${data.jobId}
209
+ Status: ${data.status}
210
+ URL: ${args.url}
211
+ Time limit: ${duration} minutes
212
+
213
+ The test has been queued for a HUMAN tester. It is NOT complete yet.
214
+
215
+ ⚠️ REQUIRED NEXT STEPS:
216
+ 1. Wait 30-60 seconds
217
+ 2. Call get_job_status(jobId: "${data.jobId}") to check if complete
218
+ 3. Repeat step 2 until status="completed" (typical time: 2-10 minutes)
219
+ 4. Call get_job_result(jobId: "${data.jobId}") to get the actual test results
220
+
221
221
  DO NOT consider this task done until you have retrieved the results!`,
222
222
  },
223
223
  ],
@@ -227,13 +227,13 @@ DO NOT consider this task done until you have retrieved the results!`,
227
227
  return {
228
228
  content: [{
229
229
  type: 'text',
230
- text: `❌ Error creating job
231
-
232
- ${error instanceof Error ? error.message : 'Unknown error'}
233
-
234
- Please check:
235
- - The API server is running at ${API_URL}
236
- - Your network connection
230
+ text: `❌ Error creating job
231
+
232
+ ${error instanceof Error ? error.message : 'Unknown error'}
233
+
234
+ Please check:
235
+ - The API server is running at ${API_URL}
236
+ - Your network connection
237
237
  - Your .env file configuration`
238
238
  }],
239
239
  isError: true
@@ -251,10 +251,10 @@ Please check:
251
251
  return {
252
252
  content: [{
253
253
  type: 'text',
254
- text: `❌ Job not found
255
-
256
- Job ID: ${args.jobId}
257
-
254
+ text: `❌ Job not found
255
+
256
+ Job ID: ${args.jobId}
257
+
258
258
  The job does not exist or you don't have permission to access it.`
259
259
  }],
260
260
  isError: true
@@ -263,9 +263,9 @@ The job does not exist or you don't have permission to access it.`
263
263
  return {
264
264
  content: [{
265
265
  type: 'text',
266
- text: `❌ Failed to get job status
267
-
268
- Status: ${response.status}
266
+ text: `❌ Failed to get job status
267
+
268
+ Status: ${response.status}
269
269
  Error: ${response.statusText}`
270
270
  }],
271
271
  isError: true
@@ -281,8 +281,8 @@ Error: ${response.statusText}`
281
281
  timeout: '⏰'
282
282
  };
283
283
  const emoji = statusEmoji[job.status] || '📊';
284
- let message = `${emoji} Job Status: ${job.status}
285
-
284
+ let message = `${emoji} Job Status: ${job.status}
285
+
286
286
  Job ID: ${job.id}`;
287
287
  if (job.status === 'pending') {
288
288
  message += '\n\nWaiting for a tester to claim this job...\nTypical completion time: 2-10 minutes';
@@ -310,8 +310,8 @@ Job ID: ${job.id}`;
310
310
  return {
311
311
  content: [{
312
312
  type: 'text',
313
- text: `❌ Error checking job status
314
-
313
+ text: `❌ Error checking job status
314
+
315
315
  ${error instanceof Error ? error.message : 'Unknown error'}`
316
316
  }],
317
317
  isError: true
@@ -329,10 +329,10 @@ ${error instanceof Error ? error.message : 'Unknown error'}`
329
329
  return {
330
330
  content: [{
331
331
  type: 'text',
332
- text: `❌ Job not found
333
-
334
- Job ID: ${args.jobId}
335
-
332
+ text: `❌ Job not found
333
+
334
+ Job ID: ${args.jobId}
335
+
336
336
  The job does not exist or you don't have permission to access it.`
337
337
  }],
338
338
  isError: true
@@ -341,9 +341,9 @@ The job does not exist or you don't have permission to access it.`
341
341
  return {
342
342
  content: [{
343
343
  type: 'text',
344
- text: `❌ Failed to get job result
345
-
346
- Status: ${response.status}
344
+ text: `❌ Failed to get job result
345
+
346
+ Status: ${response.status}
347
347
  Error: ${response.statusText}`
348
348
  }],
349
349
  isError: true
@@ -354,11 +354,11 @@ Error: ${response.statusText}`
354
354
  return {
355
355
  content: [{
356
356
  type: 'text',
357
- text: `⏳ Job not yet completed
358
-
359
- Job ID: ${job.id}
360
- Current status: ${job.status}
361
-
357
+ text: `⏳ Job not yet completed
358
+
359
+ Job ID: ${job.id}
360
+ Current status: ${job.status}
361
+
362
362
  ${job.status === 'pending' ? 'Waiting for a tester to claim this job...' :
363
363
  job.status === 'claimed' || job.status === 'in_progress' ? 'The tester is working on your test...' :
364
364
  job.status === 'failed' ? '❌ Job failed. Error: ' + (job.error || 'Unknown error') :
@@ -368,10 +368,10 @@ ${job.status === 'pending' ? 'Waiting for a tester to claim this job...' :
368
368
  };
369
369
  }
370
370
  // Job is completed, format results
371
- const message = `✅ Test completed!
372
-
373
- Job ID: ${job.id}
374
-
371
+ const message = `✅ Test completed!
372
+
373
+ Job ID: ${job.id}
374
+
375
375
  **Test Results:**`;
376
376
  const contents = [
377
377
  {
@@ -395,8 +395,8 @@ Job ID: ${job.id}
395
395
  return {
396
396
  content: [{
397
397
  type: 'text',
398
- text: `❌ Error getting job result
399
-
398
+ text: `❌ Error getting job result
399
+
400
400
  ${error instanceof Error ? error.message : 'Unknown error'}`
401
401
  }],
402
402
  isError: true
package/package.json CHANGED
@@ -1,62 +1,62 @@
1
- {
2
- "name": "@runhuman/mcp-server",
3
- "version": "1.1.0",
4
- "description": "Model Context Protocol (MCP) server for RunHuman - Human-powered QA testing for AI agents",
5
- "main": "dist/index.js",
6
- "type": "module",
7
- "bin": {
8
- "runhuman-mcp": "dist/index.js"
9
- },
10
- "files": [
11
- "dist",
12
- "README.md",
13
- ".env.example"
14
- ],
15
- "scripts": {
16
- "build": "tsc",
17
- "dev": "tsx watch src/index.ts",
18
- "start": "node dist/index.js",
19
- "prepublishOnly": "npm run build",
20
- "test": "node test-simple.cjs",
21
- "test:all": "node test-all-tools.cjs",
22
- "test:routes": "node test-route-fix.cjs",
23
- "test:endpoints": "node test-api-endpoints.cjs",
24
- "test:ci": "npm run test:endpoints && npm run test:routes",
25
- "test:inspector": "npx @modelcontextprotocol/inspector node dist/index.js"
26
- },
27
- "keywords": [
28
- "mcp",
29
- "model-context-protocol",
30
- "qa",
31
- "testing",
32
- "human-in-the-loop",
33
- "ai-agent",
34
- "claude",
35
- "anthropic",
36
- "qa-testing",
37
- "manual-testing"
38
- ],
39
- "author": "RunHuman <hey@runhuman.com>",
40
- "repository": {
41
- "type": "git",
42
- "url": "git+https://github.com/yueranyuan/qa-experiment.git",
43
- "directory": "packages/mcp-server"
44
- },
45
- "homepage": "https://runhuman.com",
46
- "bugs": {
47
- "url": "https://github.com/yueranyuan/qa-experiment/issues"
48
- },
49
- "license": "ISC",
50
- "engines": {
51
- "node": ">=18.0.0"
52
- },
53
- "dependencies": {
54
- "@modelcontextprotocol/sdk": "latest",
55
- "dotenv": "^17.2.3"
56
- },
57
- "devDependencies": {
58
- "@types/node": "^20.11.17",
59
- "tsx": "^4.7.1",
60
- "typescript": "^5.3.3"
61
- }
62
- }
1
+ {
2
+ "name": "@runhuman/mcp-server",
3
+ "version": "2.0.0",
4
+ "description": "Model Context Protocol (MCP) server for RunHuman - Human-powered QA testing for AI agents",
5
+ "main": "dist/index.js",
6
+ "type": "module",
7
+ "bin": {
8
+ "runhuman-mcp": "dist/index.js"
9
+ },
10
+ "files": [
11
+ "dist",
12
+ "README.md",
13
+ ".env.example"
14
+ ],
15
+ "scripts": {
16
+ "build": "tsc",
17
+ "dev": "tsx watch src/index.ts",
18
+ "start": "node dist/index.js",
19
+ "prepublishOnly": "npm run build",
20
+ "test": "node test-simple.cjs",
21
+ "test:all": "node test-all-tools.cjs",
22
+ "test:routes": "node test-route-fix.cjs",
23
+ "test:endpoints": "node test-api-endpoints.cjs",
24
+ "test:ci": "npm run test:endpoints && npm run test:routes",
25
+ "test:inspector": "npx @modelcontextprotocol/inspector node dist/index.js"
26
+ },
27
+ "keywords": [
28
+ "mcp",
29
+ "model-context-protocol",
30
+ "qa",
31
+ "testing",
32
+ "human-in-the-loop",
33
+ "ai-agent",
34
+ "claude",
35
+ "anthropic",
36
+ "qa-testing",
37
+ "manual-testing"
38
+ ],
39
+ "author": "RunHuman <hey@runhuman.com>",
40
+ "repository": {
41
+ "type": "git",
42
+ "url": "git+https://github.com/yueranyuan/qa-experiment.git",
43
+ "directory": "packages/mcp-server"
44
+ },
45
+ "homepage": "https://runhuman.com",
46
+ "bugs": {
47
+ "url": "https://github.com/yueranyuan/qa-experiment/issues"
48
+ },
49
+ "license": "ISC",
50
+ "engines": {
51
+ "node": ">=18.0.0"
52
+ },
53
+ "dependencies": {
54
+ "@modelcontextprotocol/sdk": "latest",
55
+ "dotenv": "^17.2.3"
56
+ },
57
+ "devDependencies": {
58
+ "@types/node": "^20.11.17",
59
+ "tsx": "^4.7.1",
60
+ "typescript": "^5.3.3"
61
+ }
62
+ }