Skip to main content

What Are Recipes?

Recipes are pre-designed combinations of connectors that solve specific use cases. Instead of figuring out which connectors to combine and how to configure them, use a recipe to get started instantly.
Think of recipes as “templates” or “blueprints” for building powerful AI agents.

Why Use Recipes?

Proven Patterns

Based on real-world production workflows used by teams

Save Time

Skip trial and error - start with working configurations

Best Practices

Optimized tool selections and permissions

Customizable

Use as starting point, adjust to your needs

Available Recipes

How to Use a Recipe

1

Choose a Recipe

Browse recipes and select one that matches your use case
2

Click 'Use This Recipe'

In the recipe page, click the “Use This Recipe” button
3

Connect Services

Authenticate required connectors (OAuth, one-click)
4

Review Tools

See which tools are enabled by default. Adjust if needed.
5

Deploy

Your toolset is deployed and ready to use!
6

Test in Playground

Try example workflows to see it in action
7

Connect to AI

Add to Claude Desktop, Cursor, or your AI agent

Recipe Components

Each recipe includes:

Connectors

List of pre-built integrations to connect:
  • Which services (Linear, GitHub, Slack, etc.)
  • Why each is needed
  • Alternative options (e.g., Jira instead of Linear)

Tools

Specific tools enabled from each connector:
  • Read operations (safe, always enabled)
  • Write operations (carefully selected)
  • Blocked tools (dangerous operations disabled by default)

Configuration

Recommended settings:
  • Tool permissions (allow, block, require approval)
  • Optimization features (knowledge indexing, code indexing)
  • Workflows (multi-tool automation sequences)

Example Workflows

Real-world scenarios showing the agent in action:
  • Step-by-step tool execution
  • Expected outcomes
  • Common edge cases

Customizing Recipes

Recipes are starting points. Customize them:
Expand the recipe with additional services:
  • Triage Agent + PagerDuty for alerting
  • Support Agent + Salesforce for CRM
  • DevOps Agent + DataDog for monitoring
Don’t need all connectors? Remove unused ones:
  • Triage Agent without Slack (if you use Discord)
  • Support Agent without Intercom (if you use Zendesk)
Change which tools are allowed:
  • Block destructive operations (delete, archive)
  • Require approval for sensitive actions (deploy, merge PR)
  • Enable additional tools as you gain confidence
Enhance with search capabilities:
  • Add company documentation (knowledge indexing)
  • Add code repositories (code indexing)
  • Agents can reference docs and code automatically

Recipe Comparison

RecipeTime to SetupComplexityPrimary Use Case
Triage Agent10 minMediumIncident response
Customer Support10 minLowTicket handling
DevOps Automation15 minHighDeployment workflows
Data Analysis10 minLowQuerying & reporting

Success Stories

Company: TechCorp (50 engineers)Before: 30 minutes to triage production errors manuallyAfter: 3 minutes with Triage AgentImpact:
  • 10x faster incident response
  • Reduced MTTR from hours to minutes
  • Engineers focus on fixes, not investigation
Company: SaaS Startup (10-person support team)Before: 40% of tickets escalated to engineeringAfter: 20% escalation rate with Support AgentImpact:
  • Agents answer common questions automatically
  • Engineering team freed up for development
  • Faster customer response times
Company: E-commerce PlatformBefore: Manual deployment process, 30 minutes per deployAfter: Automated with DevOps Agent, 5 minutesImpact:
  • 5x more frequent deployments
  • Fewer human errors
  • Automated rollback on errors

Best Practices

Don’t try to implement all recipes at once:
  1. Choose your biggest pain point
  2. Implement one recipe
  3. Test thoroughly
  4. Roll out to team
  5. Then add more recipes
Gradual adoption = higher success rate
Before deploying to production:
  1. Test each tool individually
  2. Test multi-tool workflows
  3. Test error handling
  4. Verify tool permissions
Catch issues early in safe environment
Connect test workspaces first:
  • Linear: Test workspace
  • Slack: Test channel (#testing)
  • GitHub: Test repository
Validate without affecting production.
After deploying:
  • Track tool usage
  • Monitor error rates
  • Review execution traces
  • Optimize based on data
Data-driven iteration
Document your configuration:
  • Which recipe you’re using
  • Customizations made
  • Example workflows
  • Contact for questions
Enable team adoption

Creating Your Own Recipe

Have a unique workflow? Create a custom recipe:
  1. Identify Use Case: What problem are you solving?
  2. Select Connectors: Which services do you need?
  3. Configure Tools: Enable relevant operations
  4. Test Workflows: Verify multi-tool sequences work
  5. Document: Write down the workflow steps
  6. Share: Export configuration, share with team

Need Help?

Contact support@tadata.com to discuss custom recipes for your use case.

Next Steps