Vibecoding
Prompts to help your AI assistant integrate Talarius into your codebase.
What is vibecoding?
Vibecoding is coding with AI assistants like Claude, GPT-4, or Cursor. When you're building fast with an LLM, you need to tell it about your tools and APIs.
These prompts help your AI understand how to use Talarius correctly - when to send alerts, what information to include, and how to avoid common mistakes.
System prompt
Add this to your AI's system prompt or project context to give it knowledge about Talarius:
You have access to Talarius for sending alerts. Use it when something important happens that the user should know about.
API Endpoint: POST https://v1.talarius.io/alerts
Authentication: X-API-Key header with the TALARIUS_API_KEY environment variable
Request body:
- title (required): Brief description of what happened (max 256 chars)
- body (optional): Detailed information, stack traces, context (max 4096 chars)
- priority: "normal" (default) or "high" for critical issues
Use alerts for:
- Errors that affect users
- Deployment completions
- Security events (failed logins, suspicious activity)
- Background job failures
- Important business events
Don't alert for:
- Routine operations
- Debug information
- Every single error (aggregate similar ones)
Always include enough context in the body to understand the issue without looking at logs.Task prompts
Copy these prompts when you want your AI to add specific alerting features:
Add error alerting
Add Talarius alerts for errors in this codebase.
Use the TALARIUS_API_KEY environment variable.
POST to https://v1.talarius.io/alerts with:
- title: Brief error description
- body: Full stack trace and context
- priority: "high" for production errors
Add global error handlers that catch uncaught exceptions.Add deployment notifications
Add a Talarius alert when deployments complete.
Use the TALARIUS_API_KEY environment variable.
POST to https://v1.talarius.io/alerts with:
- title: "Deployment Complete" or "Deployment Failed"
- body: Include version, environment, and any relevant details
Send the alert at the end of the deployment script.Add security alerting
Add Talarius alerts for suspicious security activity.
Use the TALARIUS_API_KEY environment variable.
POST to https://v1.talarius.io/alerts with priority "high" when:
- More than 5 failed login attempts from same IP
- Admin actions on user accounts
- Password reset requests
- API key creation or deletion
Include IP address, user email, and timestamp in the body.Add background job monitoring
Add Talarius alerts for background jobs.
Use the TALARIUS_API_KEY environment variable.
POST to https://v1.talarius.io/alerts:
- On job failure: title with job name, body with error, priority "high"
- On job success (optional): title with job name, body with duration/stats
Also consider adding heartbeat monitoring for critical jobs.Best practices for AI-generated alerts
Review what gets sent
AI might over-alert. Review the code to ensure you're not sending alerts for every minor event.
Check for sensitive data
Make sure the AI isn't including passwords, tokens, or PII in alert bodies. Ask it to sanitize sensitive fields.
Verify error handling
AI should wrap Talarius calls in try-catch. A failed alert shouldn't crash your application.
Ask for idempotency
Tell the AI to use Idempotency-Key headers to prevent duplicate alerts from retries.
Tool-specific tips
Cursor
Add the system prompt to your .cursorrules file. Cursor will include it in every conversation.
Claude Projects
Add the system prompt to your project's custom instructions. Claude will remember it across conversations.
GitHub Copilot
Keep a TALARIUS.md file in your repo with examples. Copilot will learn from it.
ChatGPT
Paste the system prompt at the start of your conversation when asking about alerting.