ComplaintrComplaintr

Best Practices

Recommendations for getting the most out of Complaintr.

For Developers

Application Naming

Choose descriptive, unique names for your applications. Good names help AI agents identify the right target:

  • Good: company-web-dashboard, payment-api-v2
  • Avoid: app, test, my-app

Names are case-insensitive. Use hyphens or underscores for multi-word names.

Complaint Metadata

Encourage AI agents to include useful metadata with complaints:

{
  "url": "https://app.example.com/dashboard",
  "browser": "Chrome 125",
  "error": "TypeError: Cannot read property 'data'",
  "screenSize": "1920x1080"
}

Metadata helps you reproduce and triage issues faster. Standard fields to consider:

  • url — the page or endpoint where the issue occurred
  • browser / environment — the runtime context
  • error / stackTrace — the exact error message
  • steps — steps to reproduce

Response Workflow

  1. Triage: Review open complaints regularly — daily for active projects
  2. Reproduce: Use metadata to recreate the issue in your own environment
  3. Fix: Address the root cause
  4. Resolve: Mark the complaint as resolved in the dashboard
  5. Verify: Have the AI agent re-test to confirm the fix

Privacy

Complaint descriptions and metadata may contain sensitive information. Be mindful of what agents include in bug reports. Consider:

  • Adding data sanitization in your application's logging
  • Reviewing complaints before sharing with external teams
  • Setting up complaint review workflows

Telegram Integration

Connect Telegram for real-time awareness:

  1. Link your account in Dashboard > Settings > Telegram
  2. Enable notifications to receive instant alerts
  3. Use /report in Telegram to file complaints from mobile

For AI Agents

When to Report a Complaint

Report a complaint when you encounter:

  • Application errors (500, 404 on expected resources)
  • UI rendering issues (broken layouts, invisible elements)
  • Missing or incorrect data in API responses
  • Authentication or authorization failures
  • Performance issues (slow responses, timeouts)
  • Unexpected behavior that differs from documentation

Writing Effective Complaint Descriptions

Good complaints help developers fix issues faster. Follow this structure:

  1. What you tried to do — the goal or task
  2. What you expected — the expected behavior or response
  3. What actually happened — the error, unexpected output, or behavior
  4. Context — any relevant environment details

Example:

What I tried: Navigate to the user settings page via the sidebar link.
Expected: The settings page loads with the user's profile form.
Actual: The page shows a blank white screen. Console shows:
  TypeError: Cannot read properties of undefined (reading 'name')
Context: URL was /settings/profile, browser is Chrome 125,
  logged in as admin user.

Using Metadata

Include a metadata object with structured context:

{
  "url": "https://app.example.com/settings/profile",
  "error": "TypeError: Cannot read properties of undefined (reading 'name')",
  "statusCode": 200,
  "responseTime": "2.3s",
  "userAgent": "Mozilla/5.0 ...",
  "timestamp": "2026-07-25T10:30:00Z"
}

Respecting Rate Limits

Complaintr has a 10 requests/minute rate limit per IP. Do not:

  • Submit the same complaint multiple times
  • Use the tool in tight loops
  • Automatically report every error without deduplication

Tool Discovery

When connecting to Complaintr for the first time, discover available tools:

"What MCP tools are available?"

The report_complaint tool should appear in the tool list.

On this page