报告门户MCP服务器

Prerequisites

Before setting up the MCP server, you need the following information from your ReportPortal instance:

Prerequisites

• Go 1.24.4 or later
• A ReportPortal instance

Dependencies

Run task deps to install Go dependencies:

task deps

Installation

There are two ways to connect to the ReportPortal MCP Server: 1. Locally - via Docker (recommended) or using pre-built binaries. 2. Connecting to a remote MCP server (when the server is already deployed)

Each of these methods is suitable for any LLM provider.

Local installation

The configurations below use the default stdio mode (MCP_MODE=stdio), which is the correct choice for all local AI tool integrations. To run the server in HTTP mode instead, add MCP_MODE=http to the env block (see the For developers section for details).

Via Docker (recommended)

The MCP server is available on the official ReportPortal's DockerHub.

Configuration:

{
  "reportportal": {
    "command": "docker",
    "args": [
      "run",
      "-i",
      "--rm",
      "-e",
      "RP_API_TOKEN",
      "-e",
      "RP_HOST",
      "-e",
      "RP_PROJECT",
      "reportportal/mcp-server"
    ],
    "env": {
      "RP_API_TOKEN": "your-api-token",
      "RP_HOST": "https://your-reportportal-instance.com",
      "RP_PROJECT": "YourProjectKeyFromReportPortal"
    }
  }
}

Using pre-built binaries

The OS pre-built binaries can be downloaded from the official releases on GitHub.

Configuration:

{
  "reportportal": {
    "command": "/path/to/reportportal-mcp-server-binary",
    "args": ["stdio"],
    "env": {
      "RP_API_TOKEN": "your-api-token",
      "RP_HOST": "https://your-reportportal-instance.com",
      "RP_PROJECT": "YourProjectKeyFromReportPortal"
    }
  }
}

AI Tool Setup

Choose your favourite AI Tool to connect.

Building from Source

```bash

Build the binary

go build -o reportportal-mcp-server ./cmd/reportportal-mcp-server ```

This creates an executable called reportportal-mcp-server.

Build

task build

Build with Docker

task docker:build

Verifying Your Setup

2. Verify Remote MCP Server (if using remote deployment)

If connecting to a remote MCP server, verify it's accessible:

Via Browser:

http://your-mcp-server-host:port/

Should return server information and available endpoints.

Via curl (GET request):

```bash

Docker Issues

Problem: Docker container exits immediately

Solutions: 1. Check container logs: docker logs <container-name> 2. Verify all required environment variables are set 3. Ensure environment variable values don't have syntax errors 4. Check Docker has permission to access the network

Problem: "docker: command not found"

Solutions: 1. Install Docker Desktop for your OS 2. Verify Docker is running: docker --version 3. For Linux, add user to docker group: sudo usermod -aG docker $USER

Example Queries (Natural Language)

Here are some real-world examples of what you might ask your AI after setup (the assistant's response will be drawn from ReportPortal data):

• "List the 5 most recent test launches." – returns a paginated list of recent test runs with names and statuses.
• "What tests failed in the latest run?" – shows failed test items for the most recent launch.
• "Show me details of launch with ID 119000." – retrieves a specific launch directly by its ID without pagination.
• "Show me details of launch with number 1234." – fetches information (ID, name, description, stats) for that specific launch.
• "Run quality gate on launch 12345." – executes quality gate analysis to verify if launch meets defined quality criteria.
• "Run an analysis on launch ABC." – triggers the ReportPortal's auto-analysis to summarize results and failures for launch "ABC".
• "Finish the running launch with ID 4321." – forces a currently running test launch to stop.
• "Show me the top five 500-level errors in the last hour" - lists the top 5 such errors from the recent test results.
• "Import this JUnit XML report into project 'my_project' using the junit plugin." – uploads the provided report file to ReportPortal and creates a new launch from it.

Each query above corresponds to a "tool" provided by the MCP server, but you just phrase it naturally. The AI will invoke the correct command behind the scenes. These features let you query and manage your test reports in many ways through simple chat interactions.

Project Key (`RP_PROJECT`) — Optional

This value is optional. When set, it defines the default project key used for all requests; individual tools can still override it per call via the projectKey argument.

The project key is the unique project identifier within the ReportPortal instance, do not use the project display name as project key. Find this on the ReportPortal general settings page:

https://your-rp-instance.com/ui/#organizations/{your-organization}/projects/{your_project}/settings/general

Configuration

The server needs to know where your ReportPortal is and how to authenticate. Set these environment variables in your shell:

For stdio mode (default):

For HTTP mode:

Set MCP_MODE=http and configure the following: - RP_HOST: Required - The URL of your ReportPortal - RP_PROJECT: Optional - Your default project key (unique project identifier within the ReportPortal instance) - MCP_SERVER_PORT: Optional - HTTP server port (default: 8080) - MCP_SERVER_HOST: Optional - HTTP bind host (default: empty) - Authentication tokens must be passed per-request via Authorization: Bearer <token> header - RP_API_TOKEN environment variable is not used in HTTP mode

Example for stdio mode:

export RP_HOST="https://your-reportportal-instance.com"
export RP_PROJECT="YourProjectKeyFromReportPortal"
export RP_API_TOKEN="your-api-token"
./reportportal-mcp-server

Example for HTTP mode:

```bash export MCP_MODE=http export RP_HOST="https://your-reportportal-instance.com" export RP_PROJECT="YourProjectKeyFromReportPortal" export MCP_SERVER_PORT=8080 ./reportportal-mcp-server

Configuration Issues

Problem: "Environment variable not set"

Solutions: 1. Verify variable names are correct: RP_HOST, RP_PROJECT, RP_API_TOKEN 2. For Docker, check -e flags are specified correctly 3. For binary, export variables in the same shell session 4. Use echo $VAR_NAME (Linux/Mac) or $env:VAR_NAME (PowerShell) to verify

Problem: "Invalid JSON" configuration errors

Solutions: 1. Validate JSON syntax using a JSON validator 2. Remove trailing commas in JSON objects 3. Ensure all strings are properly quoted 4. Check for comments (JSON doesn't support comments) 5. Verify escape characters in paths (use \\ in Windows paths)

API Token (`RP_API_TOKEN`)

You can get an API token from your ReportPortal Profile or generate a new one. Security Note: Never commit tokens to version control or share them publicly.

HTTP API Endpoints

When running in HTTP mode (MCP_MODE=http), the server exposes the following endpoints:

MCP Endpoints (for tool calls and MCP protocol)

• POST /mcp - Main MCP endpoint for JSON-RPC requests
• POST /api/mcp - Alternative MCP endpoint (same functionality)
• GET /mcp - SSE (Server-Sent Events) stream endpoint for MCP protocol
• GET /api/mcp - Alternative SSE stream endpoint

Important: POST requests must be sent to /mcp or /api/mcp, not to the root endpoint /.

Request Format:

All MCP requests must follow the JSON-RPC 2.0 specification:

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "id": 1,
  "params": {
    "name": "get_launches",
    "arguments": {
      "filter-cnt-name": "test",
      "page": 1,
      "page-size": 10
    }
  }
}

Example Request:

curl -X POST http://localhost:8080/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer your-reportportal-token" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "id": 1,
    "params": {
      "name": "get_launches",
      "arguments": {
        "page": 1,
        "page-size": 10
      }
    }
  }'

Information Endpoints (GET only)

• GET / - Root endpoint, returns server information and available endpoints
• GET /health - Health check endpoint
• GET /info - Server information and configuration
• GET /api/status - Server status (same as /info)
• GET /metrics - Analytics metrics (if analytics enabled)

Note: The root endpoint / only accepts GET requests. POST requests to / will return a 404 error. Use /mcp or /api/mcp for MCP protocol requests.

Test API access with your token

curl -H "Authorization: Bearer your-api-token" \ https://your-reportportal-instance.com/api/v1/YourProjectKey/launch

**Via PowerShell:**

Test API access

$headers = @{ "Authorization" = "Bearer your-api-token" } Invoke-RestMethod -Uri "https://your-reportportal-instance.com/api/v1/YourProjectKey/launch" -Headers $headers ```

Expected Results: - HTTP 200 OK response - Valid JSON response with launch data (if project has launches) - No authentication errors (401) or forbidden errors (403)

Test MCP endpoint with JSON-RPC request

curl -X POST http://your-mcp-server-host:port/mcp \ -H "Content-Type: application/json" \ -H "Authorization: Bearer your-api-token" \ -H "X-Project: YourProjectKey" \ -d '{ "jsonrpc": "2.0", "method": "tools/list", "id": 1 }'

**Via PowerShell:**

Test MCP endpoint

$headers = @{ "Content-Type" = "application/json" "Authorization" = "Bearer your-api-token" "X-Project" = "YourProjectKey" } $body = @{ jsonrpc = "2.0" method = "tools/list" id = 1 } | ConvertTo-Json

Invoke-RestMethod -Uri "http://your-mcp-server-host:port/mcp" -Method Post -Headers $headers -Body $body

**Via ping (network connectivity only):**

Expected Results: - Server responds to health checks - /info returns server configuration - MCP endpoint returns list of available tools - No connection refused or timeout errors

3. Verify MCP Server Integration

After configuration, verify the AI assistant can communicate with the MCP server:

Step 1: Check Available Tools

Ask your AI assistant:

"What ReportPortal tools are available?"

Expected response: A list of 31 tools including launches, test items, analysis tools, TMS tools, etc.

Step 2: Test Basic Query

Try a simple query:

"List the 5 most recent test launches"

Expected response: A formatted list of recent launches with names, statuses, and timestamps.

Step 3: Check Server Logs

Monitor logs to verify requests are being processed:

Docker:

```bash

AI Assistant Integration Issues

Problem: AI assistant doesn't recognize MCP server

Solutions: 1. Verify configuration file syntax is valid JSON 2. Check configuration file is in the correct location 3. Restart the AI assistant after configuration changes 4. For Cursor/VS Code, check MCP extension is installed and enabled 5. Review AI assistant logs for error messages

Problem: Tools list is empty

Solutions: 1. Verify MCP server is running and accessible 2. Check server logs for startup errors 3. Ensure server mode matches client configuration (stdio vs HTTP) 4. For remote servers, verify URL ends with /mcp/

GitHub Copilot (In VS Code and JetBrains IDEs)

VS Code

1. Install/update the GitHub Copilot plugin.
2. Press Ctrl+P and type >mcp in the search bar and select MCP: Open User Configuration.
3. That will open file mcp.json where you need to add a new MCP server entry that runs the ReportPortal MCP Server:

For local installation (Docker or binary):

Choose your preferred configuration from the Installation section and paste it inside the reportportal block.

{
  "servers": {
    "reportportal": {
      // paste your chosen configuration here
    }
  }
}

For remote server:

Note: The remote server must be deployed and running in HTTP mode before connecting.

{
  "servers": {
    "reportportal": {
      "url": "http://your-mcp-server-host:port/mcp/",
      "requestInit": {
        "headers": {
          "Authorization": "Bearer ${RP_API_TOKEN}",
          "X-Project": "YourProjectKeyFromReportPortal"
        }
      }
    }
  }
}

Documentation: VS Code Copilot Guide.

JetBrains IDEs

1. Install/update the GitHub Copilot plugin.
2. Click GitHub Copilot icon in the status bar → Edit Settings → Model Context Protocol → Configure.
3. Add configuration:

For local installation (Docker or binary):

Choose your preferred configuration from the Installation section and paste it inside the reportportal block.

{
  "servers": {
    "reportportal": {
      // paste your chosen configuration here
    }
  }
}

For remote server:

Note: The remote server must be deployed and running in HTTP mode before connecting.

{
  "servers": {
    "reportportal": {
      "url": "http://your-mcp-server-host:port/mcp/",
      "requestInit": {
        "headers": {
          "Authorization": "Bearer ${RP_API_TOKEN}",
          "X-Project": "YourProjectKeyFromReportPortal"
        }
      }
    }
  }
}

1. Press Ctrl + S or Command + S to save, or close the mcp.json file. The configuration should take effect immediately and restart all the MCP servers defined. You can restart the IDE if needed.

Documentation: JetBrains Copilot Guide.

Troubleshooting