Test Mcp Connector
Test the MCP connector for provider: $ARGUMENTS
Test the MCP connector for provider: $ARGUMENTS
## CRITICAL: Progress Communication
**You MUST provide frequent status updates throughout this process.** Testing can take 5-10 minutes, and users need visibility into what's happening. Do NOT run tests silently in the background.
## Required Setup
Gather from user:
- StackOne account ID (e.g., Wc8pkmCj0HaVTGnWbWC12)
- Connector/provider name (e.g., bamboohr, datadog)
- StackOne API key for MCP server authentication
- CLI profile for pushing connector to StackOne
## Execution Phases
### Phase 0: Progress Tracking Setup
**IMMEDIATELY** create a todo list to show the user what's happening:
Use TodoWrite with:
- Read connector configuration files
- Identify all testable actions
- Set up test environment
- Run tests with progress reporting
- Generate final report
Update todo status as you complete each phase.
### Phase 1: Read Configuration
1. Search for connector files: `connectors/[provider]/*.yaml`
2. Read main connector and all partial files
3. **Output to user**:
📂 Reading connector configuration files... 📋 Found [N] actions across [M] config files Categories: [list categories like employees, time_off, etc.]
### Phase 2: Environment Setup
1. Create test directory, init npm, install dependencies
2. **Output to user**: "🔧 Setting up test environment..."
3. **Output when ready**: "🔧 Test environment ready"
### Phase 3: Generate Test Script
Generate test script with **BATCHED EXECUTION** for progress visibility:
```typescript
const BATCH_SIZE = 10; // Report progress every 10 tests
// After each batch, output:
console.log(`⏳ Progress: ${completed}/${total} (${pct}%) - ✅ ${passed} passed, ❌ ${failed} failed`);
Phase 4: Run Tests
IMPORTANT: Run tests in FOREGROUND (not background) so output streams live.
🧪 Starting test run - this typically takes 5-7 minutes for large connectors...
⏳ Progress: 10/99 (10%) - ✅ 10 passed, ❌ 0 failed
⏳ Progress: 20/99 (20%) - ✅ 19 passed, ❌ 1 failed
...
If you MUST use background execution:
- Poll every 30-60 seconds with
TaskOutput(block: false) - Output: "⏳ Tests still running... [elapsed time]"
- Continue until complete
Phase 5: Report Results
============================================================
Test Results: [provider] Connector
============================================================
Total: [N] | Passed: [X] ✅ | Failed: [Y] ❌ | Pass Rate: [%]%
Failed Actions (if any):
- [action]: [error]
============================================================
Phase 6: Fix Loop (if failures)
When issues found:
- Diagnose root cause from error messages
- Fix connector YAML
- Run
stackone pushto deploy fix - Retry failed tests
Test Approach
Quick Connectivity Check (optional): Run
stackone runto verify basic connectivityAgent Simulation (primary): Build TypeScript test harness using Claude Agent SDK with Haiku:
- Generate domain-aware test scenarios (not mechanical transforms)
- Test if agents can discover and correctly use actions
- Capture agent reasoning to diagnose issues
- Show conversations and progress as tests run
Anti-Patterns (NEVER DO THESE)
- ❌ Running tests in background without progress polling
- ❌ Staying silent for more than 60 seconds during test execution
- ❌ Outputting results only at the very end
- ❌ Not using TodoWrite to show phases
- ❌ Not telling user how long tests typically take
Example Output Flow
🔍 Starting connector test for bamboohr...
📂 Reading connector configuration files...
📋 Found 99 actions across 11 config files
Categories: employees, time_off, benefits, goals, time_tracking, files, training, applicant_tracking, meta, reports, webhooks
🔧 Setting up test environment...
🔧 Test environment ready
🧪 Starting test run - this typically takes 5-7 minutes for large connectors...
⏳ Progress: 10/99 (10%) - ✅ 10 passed, ❌ 0 failed
⏳ Progress: 20/99 (20%) - ✅ 19 passed, ❌ 1 failed
⏳ Progress: 30/99 (30%) - ✅ 28 passed, ❌ 2 failed
⏳ Progress: 40/99 (40%) - ✅ 38 passed, ❌ 2 failed
...
⏳ Progress: 99/99 (100%) - ✅ 95 passed, ❌ 4 failed
============================================================
Test Results: bamboohr Connector
============================================================
Total: 99 | Passed: 95 ✅ | Failed: 4 ❌ | Pass Rate: 95.9%
Failed Actions:
- create_webhook: 403 Forbidden - insufficient permissions
- delete_employee: 404 Not Found - test data not found
============================================================
Usage: /test-mcp-connector
Maintain Test Mcp Connector?
Let people know it's listed here — add the badge (live metrics, light/dark aware) or a plain link to your README or docs.
[Test Mcp Connector on getagentictools](https://getagentictools.com/loops/stackonehq-test-mcp-connector?ref=badge)