Configure your first agent

What is an agent?

An agent is a configurable AI endpoint with:

• a stable slug used as model
• a routing strategy
• a system prompt
• optional knowledge bases and MCP tools

1) Create the agent

1. Dashboard → Agents → New Agent
2. Set a name
3. Set a slug (used in API requests)
4. Save

2) Define system prompt

Use clear role, boundaries, format, and tone instructions.

3) Choose routing

• Fixed: one deployment for all requests
• Rule-Based: route by conditions
• Workflow: graph editor for advanced pipelines

4) Connect knowledge base (optional)

Enable one or more knowledge bases in Knowledge & Tools.

5) Test via API

Create an API key and call:

curl https://your-domain.com/api/proxy/v1/chat/completions \
  -H "Authorization: Bearer myk_..." \
  -H "Content-Type: application/json" \
  -d '{"model":"support-agent","messages":[{"role":"user","content":"Hello"}]}'

Next steps

• Set up RAG pipeline
• MCP tools guide
• API reference